R.O.I.

[bugboy]

I.m sure we have some very very successful 'capers on this board.
However, can someone put one of those vote thingys on here, I'm very courious of what king of ROI 's we have here. No names Just say maybe
.50-.60 .61-.80 .81 to 1.00 and so on. If someone would do this, sounds interesting. I don't know how to do those pool things. So if someone would, I thank you.

Frank

[Some_One]

why does it matter?

[green80]

I dont think anyone here has a 50% or better roi long term. I think you should start at 0-5%

[raybo]

I dont think anyone here has a 50% or better roi long term. I think you should start at 0-5%

So, you're saying that everyone here loses 50% of their wager costs long term? Man, why would anyone play the horses, so they can lose money? There are many more enjoyable ways of losing money, wouldn't you agree? If I lost any percentage long term I would not be playing. I also wouldn't be playing if I was only showing a couple of percent profit, it just wouldn't be worth the time and effort.

[traynor]

It might also be interesting to discover the time frame during which the ROI is calculated. A calculation made after a substantial win may be quite different than the same calculation made after a (long) string of losses. The reporting period may be a more significant element than the reported results.

One of the greatest myths in racing is the notion of ROI as equivalent to exponential growth. Anyone who believes that a "positive ROI" over some period inevitably leads to an identical (or similar) ROI over another equivalent period (other than in wagers made on paper) probably needs to get out of the house more.

Rather than asking, "how big is your ROI?" it might be interesting to discover how that ROI is calculated. Every Monday morning for the past week? The first of every month for the past month? January 1 for the past year? Or at some arbitrary time (almost invariably after a large win) that shows a more favorable result than more specific metrics applied rigorously would indicate?

http://faculty.washington.edu/chudler/stat3.html

[Phantombridgejumpe]

drink, don't smoke, never done a drug, not into music and not into reading.

I like sports and math and gambling and being entertained.

This is my hobby, and I would, in theory, love to win some money, but it
isn't my number one reason for playing.

I honestly have not made an overall profit. In fact, I'm probably down about $80-$120K over the years.

That being said, today I stopped into the OTB on a whim with $45 and walked out with about $250. I had 90 minutes of ups and downs watching about 25 races and enjoying some tremendous athletes and animals. Compared to some other hobbies, if I am indeed down $100K I consider it a bargain.

[raybo]

If one has a proper record keeping method, the ROI is calculated, and updated, daily, along with all the other important wagering metrics. If one does not have a proper record keeping method then one probably has no idea what the ROI, as well as he other metrics, is/are. The latter is probably rampant here, and everywhere else.

[Dave Schwartz]

For several years I have bought into the idea that a reasonable expectation was about 3% (before rebate).

(I think Barry Meadow's quote about that may have influenced me.)

In the last 6 months I have been mentored by a very successful player. I now believe... I almost have a hard time saying this publicly... I believe that 15-20% is possible before rebate, just as it was back in the early 90s.

Before anyone jumps on me to show proof, or accuse me of just bragging, I can only tell you that this has been my relatively recent experience over hundreds of races.

Truthfully, I am not so convinced that it has much to do with my system of handicapping or even my software. Sure, I like them both, but I have come to believe that the primary difference is my attitude.

When I say "attitude," I am referring to several key points:


Conviction that my system of handicapping is good enough to succeed without any more tinkering. (I do have a tendency to tinker with everything.)

Taking the time to paint a clear picture (of my perception) of each race.

Correctly determining which races should be strongly bet and which should be bet weakly (or not at all).

Correctly managing the money between the races (i.e. determining how much money is available to play a given race based upon current session bankroll).

Sufficient Development of courage to make the play that the system calls for despite not having lots of confidence in the outcome. (The most profitable plays rarely look much like sure things.)

[raybo]

For several years I have bought into the idea that a reasonable expectation was about 3% (before rebate).

(I think Barry Meadow's quote about that may have influenced me.)

In the last 6 months I have been mentored by a very successful player. I now believe... I almost have a hard time saying this publicly... I believe that 15-20% is possible before rebate, just as it was back in the early 90s.

Before anyone jumps on me to show proof, or accuse me of just bragging, I can only tell you that this has been my relatively recent experience over hundreds of races.

Truthfully, I am not so convinced that it has much to do with my system of handicapping or even my software. Sure, I like them both, but I have come to believe that the primary difference is my attitude.

When I say "attitude," I am referring to several key points:


Conviction that my system of handicapping is good enough to succeed without any more tinkering. (I do have a tendency to tinker with everything.)

Taking the time to paint a clear picture (of my perception) of each race.

Correctly determining which races should be strongly bet and which should be bet weakly (or not at all).

Correctly managing the money between the races (i.e. determining how much money is available to play a given race based upon current session bankroll).

Sufficient Development of courage to make the play that the system calls for despite not having lots of confidence in the outcome. (The most profitable plays rarely look much like sure things.)


You won't get any flak from me, that's for sure. If one is win betting, and playing into their strengths, and very disciplined and consistent, and keeps very good records, and controls themselves, in all aspects, 1.15 - 1.25 is very achievable. If one is an exotics player, 1.40 or higher is possible, over the long term.

Anything less than about 20% would not interest me, the variance is just too high, so the inevitable downturns, with a lesser ROI than about 1.20 would soon turn me away from the game.

[DeltaLover]

Another thread revolving around ROI..

I think it is one of the most misunderstood and abused factors in horse betting.

As a performance index ROI represents a very poor metric that can easily lead to bogus conclusions.

There is nothing to be necessarily gained by having a high ROI. As horse bettors, we should not be focused, on some theoretical reward but for real and green dollars instead.

It is quite possible for a 1.50 ROI (winning 50%) strategy, to show much less return than a 1.03 strategy, if the later is finding many more bets allowing a much larger amount to be in action than the former one.

This happens to be the case in most of the cases.

A conservative model, naturally will have higher ROI, but much less plays, ending up with an inferior return to a more aggressive, lower ROI alternative who will end up with multiple profits.

The correct question to ask, is not about ROI, but final PNL given a specific bankroll. To be more precise, we need to also consider drawdown as another important characteristic of a betting model and most of the fitness functions I use these days, derive from this general rule:

FITNESS = PNL / MAX_DRAWDOWN

adding some factors and exponentials but always trying to express such a type of relation between these two metrics. The best results, at least on my simulators, appear to be in the range of 1.02 to 1.04 ROI (which does not become a factor at no point during the optimization process).

[thaskalos]

For several years I have bought into the idea that a reasonable expectation was about 3% (before rebate).

(I think Barry Meadow's quote about that may have influenced me.)

In the last 6 months I have been mentored by a very successful player. I now believe... I almost have a hard time saying this publicly... I believe that 15-20% is possible before rebate, just as it was back in the early 90s.

Before anyone jumps on me to show proof, or accuse me of just bragging, I can only tell you that this has been my relatively recent experience over hundreds of races.

Truthfully, I am not so convinced that it has much to do with my system of handicapping or even my software. Sure, I like them both, but I have come to believe that the primary difference is my attitude.

When I say "attitude," I am referring to several key points:


Conviction that my system of handicapping is good enough to succeed without any more tinkering. (I do have a tendency to tinker with everything.)

Taking the time to paint a clear picture (of my perception) of each race.

Correctly determining which races should be strongly bet and which should be bet weakly (or not at all).

Correctly managing the money between the races (i.e. determining how much money is available to play a given race based upon current session bankroll).

Sufficient Development of courage to make the play that the system calls for despite not having lots of confidence in the outcome. (The most profitable plays rarely look much like sure things.)


No need to show me any proof, Dave...I believe you.

I do have a question though:

How many races a day do you bet on while registering this 15-20% ROI?

[raybo]

Another thread revolving around ROI..

I think it is one of the most misunderstood and abused factors in horse betting.

As a performance index ROI represents a very poor metric that can easily lead to bogus conclusions.

There is nothing to be necessarily gained by having a high ROI. As horse bettors, we should not be focused, on some theoretical reward but for real and green dollars instead.

It is quite possible for a 1.50 ROI (winning 50%) strategy, to show much less return than a 1.03 strategy, if the later is finding many more bets allowing a much larger amount to be in action than the former one.

This happens to be the case in most of the cases.

A conservative model, naturally will have higher ROI, but much less plays, ending up with an inferior return to a more aggressive, lower ROI alternative who will end up with multiple profits.

The correct question to ask, is not about ROI, but final PNL given a specific bankroll. To be more precise, we need to also consider drawdown as another important characteristic of a betting model and most of the fitness functions I use these days, derive from this general rule:

FITNESS = PNL / MAX_DRAWDOWN

adding some factors and exponentials but always trying to express such a type of relation among these two metrics. The best results, at least on my simulators, appear to be in the range of 1.02 to 1.04 ROI (which does not become a factor at no point during the optimization process).

At present, in my win betting, I'm playing about 50% of the races at the tracks I play, and my ROIs at each of those tracks are all over 1.20. I usually play 2 or 3 meets at a time. So, ROI is a very good measure of where I'm at, at all times. Hit rates are only a measure of consistency for me, a smoother or rougher ride, if you will.

Sure someone playing a couple of bets a day is not going to care much about ROI, only the money in their pocket or betting account when they leave the track, or turn off the computer. But, for someone playing many races daily, ROI is a great way to measure how one is doing. All of it is dependent on the player's goals and comfort level, some will risk large amounts of cash on a small number individual plays, others will risk smaller amounts on a larger number of plays. It's all relative, and it's all specific to the individual. Some here think that the only way to measure success is by making huge profits, but they are only speaking from their own point of view, while for others, smaller, more consistent profits are what it's all about.

If one bets many hundreds of thousands of dollars per year, 1.02-3% ROIs are great, for someone who bets a few hundred, or a few thousand a year, 1.02-3% ROI wouldn't mean much to anyone interested only in making huge profits, you could do better with other types of investments. Now, make it 1.20+% ROIs instead, and it's a different ballgame, entirely. So, ROI can be exactly what one needs to concentrate on, depending on where you're coming from and where you're trying to get to. Like I said, it's all relative.

[DeltaLover]

At present, in my win betting, I'm playing about 50% of the races at the tracks I play, and my ROIs at each of those tracks are all over 1.20. I usually play 2 or 3 meets at a time. So, ROI is a very good measure of where I'm at, at all times. Hit rates are only a measure of consistency for me, a smoother or rougher ride, if you will.



I am not one to argue and I normally overlook discussion of this type but I have
spend over 30 years analyzing and betting on this game and it intrigues me when
someones else's experiences differ so greatly from mine, all I know is that if I
have had results, even closely resembling yours I would bet a lot more than six
dollars win-place-show on my horses.

[Robert Fischer]

Rather than ROI, What I try to do is increase my bankroll by 2% every day.

[Dave Schwartz]

How many races a day do you bet on while registering this 15-20% ROI?

About 4.8 per hour of live play.


FITNESS = PNL / MAX_DRAWDOWN

What is "PNL?"


I am not one to argue and I normally overlook discussion of this type but I have spend over 30 years analyzing and betting on this game and it intrigues me when someones else's experiences differ so greatly from mine, all I know is that if I have had results, even closely resembling yours I would bet a lot more than six dollars win-place-show on my horses.

Delta, these have been my feelings exactly for the last 20 years; ever since the Beyer number went into the DRF.

The mentoring I received from someone who was doing exactly that - what what was inconprehensible to me - changed all of that.

[classhandicapper]

Assuming you are skilled enough to win, to a large extent your ROI will be a function of the percentage of races you play.

If you handicap 100 races day and make 4 plays, you should be able to do better than a guy that handicaps 10 races a day and makes 1 play.

If you handicap 20 races a day and make 5-10 plays, your ROI will be lower.

Of course, the ultimate goal is profit not ROI. You may be better off with a lower ROI and more plays.

Then again, I've always contended that most people make way too many plays. IMO most of the profit is located in a subset of their plays. Still others that are currently not profitable could probably become so if they were more selective.

[CincyHorseplayer]

So, you're saying that everyone here loses 50% of their wager costs long term? Man, why would anyone play the horses, so they can lose money? There are many more enjoyable ways of losing money, wouldn't you agree? If I lost any percentage long term I would not be playing. I also wouldn't be playing if I was only showing a couple of percent profit, it just wouldn't be worth the time and effort.

I know man!Jesus,I look forward to and get excited about how I'm going to win.That's what I think about every single time I play.Break even or total loss on a single day isn't $hit.We know the heights we are ascending to bro.

[DeltaLover]

About 4.8 per hour of live play.
What is "PNL?"


Profit and Loss

Of course in the fitness fuction we can also use the final size of the bankroll as oppposed to its PNL (since the PNL can be negative it can make some calculations regarding fitness comparisons difficult to tackle)

[CincyHorseplayer]

It might also be interesting to discover the time frame during which the ROI is calculated. A calculation made after a substantial win may be quite different than the same calculation made after a (long) string of losses. The reporting period may be a more significant element than the reported results.

One of the greatest myths in racing is the notion of ROI as equivalent to exponential growth. Anyone who believes that a "positive ROI" over some period inevitably leads to an identical (or similar) ROI over another equivalent period (other than in wagers made on paper) probably needs to get out of the house more.

Rather than asking, "how big is your ROI?" it might be interesting to discover how that ROI is calculated. Every Monday morning for the past week? The first of every month for the past month? January 1 for the past year? Or at some arbitrary time (almost invariably after a large win) that shows a more favorable result than more specific metrics applied rigorously would indicate?

http://faculty.washington.edu/chudler/stat3.html

Traynor,I love ya brother but this is complete bullshit for some of us.I know what you say is absolute truth to the target audience that think and function in the specified manner but the ROI as relates to the entirety of a single person's total game is wrong.

Your exponential growth slash negation of positive expectation theory is a complete and total disease of thinking.I know we shouldn't mention numbers on here at all because it's too much BS able,but I was completely pissed off I had an 8% ROI last year,when I had over a 30% profit on win/place bets.With a high cash rate on exotic bets and a high return when you do score the ROI should be embarrassingly 40% and should be closer to 60% IMO.I was embarrassed.That was my forte when I arrived here in 2008.Long story short I just don't see how expectations are so lowered to think that it's unduplicable unless you look at the guy in the mirror and think he's all wrong.I have won and lost and lost plenty and will lose plenty.But have won overall and expect to.I like to bet things people hate so I don't expect to lose an edge.Anyway argghhhh!You are a completely smart individual that is one of the rare influences on me,don't type out this disease!!

[Dave Schwartz]

I thought that might be it, but I always saw "P&L."

Not sure that I like your formula without a time component, but I get where you are going with it.

:ThmbUp:

Of course, with the direction I have chosen, ROI is not the goal, but an indicator that the real goal is even possible.

With...

* 4-5 plays per hour (during prime racing time)
* 15% ROI
* high hit rate (currently around 60% of races)

... the sky SHOULD be the limit. We'll see.

I can tell you that it is very fatiguing.

[CincyHorseplayer]

Return On Investment.Your 4% on betting a million is my 40% on betting 100K.Don't demean it money bags!Lot more rich and a lot more stupid.Is that a paid off virtue?!I see threads like this all the time.Bet gabs of money to take a slim slice and feel omnipotent by demeaning the guy that bets 10 times less money and comes out with the same take.I think it's arrogant.Didn't you say you were a theorist anyway?idealize a millionaire's money while sweating a pauper's bankroll?


Another thread revolving around ROI..

I think it is one of the most misunderstood and abused factors in horse betting.

As a performance index ROI represents a very poor metric that can easily lead to bogus conclusions.

There is nothing to be necessarily gained by having a high ROI. As horse bettors, we should not be focused, on some theoretical reward but for real and green dollars instead.

It is quite possible for a 1.50 ROI (winning 50%) strategy, to show much less return than a 1.03 strategy, if the later is finding many more bets allowing a much larger amount to be in action than the former one.

This happens to be the case in most of the cases.

A conservative model, naturally will have higher ROI, but much less plays, ending up with an inferior return to a more aggressive, lower ROI alternative who will end up with multiple profits.

The correct question to ask, is not about ROI, but final PNL given a specific bankroll. To be more precise, we need to also consider drawdown as another important characteristic of a betting model and most of the fitness functions I use these days, derive from this general rule:

FITNESS = PNL / MAX_DRAWDOWN

adding some factors and exponentials but always trying to express such a type of relation between these two metrics. The best results, at least on my simulators, appear to be in the range of 1.02 to 1.04 ROI (which does not become a factor at no point during the optimization process).

[Robert Fischer]

ROI is fun to calculate, but you are either winning or you aren't.

Your ROI should start out a little lower, then as your bankroll grows = your ROI will be a little higher because you can now incorporate exotics. Then as your bankroll grows still more, your ROI will nosedive as rebates become a focus of your play.

Through it all, the point is bankroll growth.

[DeltaLover]

Not sure that I like your formula without a time component, but I get where you are going with it.



Time as several other factors can surely be part of the fitness function which
is the most difficult component of an optimizer to get it right. The formula I
gave just shows the fundamental rule that is subject to a lot of tweaking making
it subject to a long series of trial and error until you reach some satisfactory
results. I have experimented with many approaches trying to discover the
ultimate fitness, including moving averages, Fibonacci moves, even Fourier
analysis at some point. What seems to be a great complication on this approach
is the fact that the input vector can differ greatly from race, something that
can only be solved by clusterization of the races which requires extremely large
dataset in order to function properly.

As far as the time dimension goes, I think that it is important not only for
fitness calculations but for handicapping decisions, since the behavior of the
crowd seems to be changing very quick and seasoning it appears to be the best
approach. This means that a betting model should be able to alter itself based
in the most recent crowd selections, an approach that becomes even more
complicated if we consider how different the crowd can behave based on factors
like the circuit or the time of the year.

I certainly respect your new findings as you describe them in the previous
posting, but I still believe that the ultimate approach to the game would have
been a completely automated system that is able to make betting decisions
without any human intervention. Of course after years of related research and
development, I cannot claim that I have reached this goal (I mean YET :) )

[raybo]

I am not one to argue and I normally overlook discussion of this type but I have
spend over 30 years analyzing and betting on this game and it intrigues me when
someones else's experiences differ so greatly from mine, all I know is that if I
have had results, even closely resembling yours I would bet a lot more than six
dollars win-place-show on my horses.

Who indicated that $6 wps is an accepted truth? I'm sure I'm not the only one here who bets a percentage of operating bankroll, not some arbitrary set amount. And, I certainly would not be betting wps either, that diminishes your ROI versus win only.

ROI is an indication of success. Hit rate is an indication of consistency and ability to carry the ROI into the future. Bankroll growth is a function of bet amount and ROI and number of plays. Etc., etc., etc..

There are many metrics involved in thorough record keeping. The ones that mean the most to the individual are the ones that tell them what they wish to know. I suggest that everybody step outside the box occasionally and view things from other perspectives, you might be amazed at what you find.

As Dave stated, doing something for 20-30 years, thinking that it is correct all that time, can be very damaging in it's totality. Sometimes we just need to determine where we want to go, realistically, and then look at all the options for doing so. Sadly, many are taught, and buy into things, early on in their life/career, only to end up on the wrong end of the stick many years later, because they were too stubborn, or too lazy, to explore all options and avenues available to them during the journey. Very sad, indeed!

[Dave Schwartz]

I certainly respect your new findings as you describe them in the previous posting, but I still believe that the ultimate approach to the game would have been a completely automated system that is able to make betting decisions without any human intervention. Of course after years of related research and development, I cannot claim that I have reached this goal (I mean YET )

Yes, that is very close to what I wanted as well, although I never expected complete automation. I will always demand absolutely zero subjectivity on the part of the handicapper. I am quite happy with semi-automation (i.e. user babysitting).

I eventually succumbed to just being (potentially) hugely profitable. If the whole process works it means settling for only a few hundred thousand per year instead of the million$.

[raybo]

This means that a betting model should be able to alter itself based
in the most recent crowd selections, an approach that becomes even more
complicated if we consider how different the crowd can behave based on factors
like the circuit or the time of the year.

I certainly respect your new findings as you describe them in the previous
posting, but I still believe that the ultimate approach to the game would have
been a completely automated system that is able to make betting decisions
without any human intervention. Of course after years of related research and
development, I cannot claim that I have reached this goal (I mean YET :) )

Now you're talking! Temper all that "old thinking" with present reality. This is the ultimate method/approach, and absolutely achievable. Incorporate recent activities, by the public and ones personal method, into what has taken place in the past, expunging the very distant past incrementally, oldest first, basing future play on the updated results and indications of future trends and indications. Automate the analysis, wagering processes, and the record keeping activity, to perform updates in real time, or near real time, and also to remove the possibility of human tendencies negatively impacting what the total automated methodology is telling you. Absolute truth is absolute truth, but that truth constantly evolves, which means the total method/approach must constantly evolve, also, in order to continue to keep the truth in sight.

[classhandicapper]

Return On Investment.Your 4% on betting a million is my 40% on betting 100K. I see threads like this all the time.Bet gabs of money to take a slim slice and feel omnipotent by demeaning the guy that bets 10 times less money and comes out with the same take.

You are describing the point I was making in my earlier post.

It's quite possible the million dollar bettor had an ROI of 0% on 900k of his total handle and 40% on his best bets or 100K of his handle and netted out to 4%.

I think it's fairly common for people to make many bets on which the expectation is negative or break even, but they don't know it. So their handle is higher, but the ROI suffers and the net profit is a non event.

The idea is not to maximize either handle or ROI. The idea is to only make bets with a profitable expectation. Most people (including myself) have a tough time finding that line and bet either too many or too few races. I have historically probably erred on the side of too few bets, but most people err on the side of too many bets.

[Dave Schwartz]

The idea is not to maximize either handle or ROI. The idea is to only make bets with a profitable expectation. Most people (including myself) have a tough time finding that line and bet either too many or too few races.

Sorry, but I would have to disagree with this.

In fact, my idea would be precisely to maximize both ROI and handle.

[classhandicapper]

Sorry, but I would have to disagree with this.

In fact, my idea would be precisely to maximize both ROI and handle.


"The idea is not to maximize either handle or ROI. The idea is to only make bets with a profitable expectation. Most people (including myself) have a tough time finding that line and bet either too many or too few races. "

If you only bet races with a positive expectation you will maximize handle and ROI.

Point being that some people bet a TON of money with a mildly negative expectation, but they don't know it because it's part of a larger group of bets that on a net basis was profitable.

Imagine this.

I get 2-1 on 10 coin flips.
I get 4-5 on 10 coin flips.

The net of the entire experience is profitable, but I'd be way better off not playing those 4-5 shots. The problem is you have to actually recognize which were which. It's my contention that most of the people that shove millions through the widows are betting those 4-5 shots a lot of the time and just don't know it.

[raybo]

Yes, that is very close to what I wanted as well, although I never expected complete automation. I will always demand absolutely zero subjectivity on the part of the handicapper. I am quite happy with semi-automation (i.e. user babysitting).

I eventually succumbed to just being (potentially) hugely profitable. If the whole process works it means settling for only a few hundred thousand per year instead of the million$.

Very proper and realistic anticipations (and potential expectations) of success! Automated/mechanical doesn't mean the absence of human intervention. After all, that's why we accomplish record keeping and stats, in order to allow for our human brains to step outside the automation and view possibilities that the automation can not see. If the automation is solid, most of the time our intervention is not needed, but from time to time, bits of data do suggest impending trends, and allow us to explore those possibilities in time to capitalize on them, before the trend has run its course. "User babysitting" is, fortunately, an advantage rather than a disadvantage, as it insures the viability of the current, and near future, automated/mechanical method.

Lowering ones "unreasonably lofty" goals is prudent, as reality proves to us what is indeed possible and achievable, and what is not. Nothing wrong with facing reality and adjusting our goals, whether upwards or downwards, as time goes by, and reality stares us in the face. :ThmbUp: :ThmbUp:

[raybo]

Sorry, but I would have to disagree with this.

In fact, my idea would be precisely to maximize both ROI and handle.

For maximum bankroll growth, this is exactly true! I think what classhandicapper is saying is that there are segments of our total play that negatively impact ROI and handle and of course bankroll growth. Of course, you already know this and I am supremely confident that your record keeping has, and is, evaluating the totality of your play to feret out those negative segments.

As to the maximization of handle, that is a function of the individual's capacity for performing work, and their desire to do so. For me at least, I don't mind performing work, but I do not wish to work myself to death either! So, my handle maximization "index" is without a doubt much lower than yours, or some others here. But, if you're willing to do what it takes, and suffer what it entails, then by all means, do it!!

[classhandicapper]

I think what classhandicapper is saying is that there are segments of our total play that negatively impact ROI and handle and of course bankroll growth. Of course, you already know this and I am supremely confident that your record keeping has, and is, evaluating the totality of your play to feret out those negative segments.



Yes. That's what's I'm saying. I think it's easier said than done when you are talking about whether something is 3%-4% profitable or 3%-4% negative.

I'll be the first to admit I am way too conservative. My typical thinking usually goes something like this.

I've narrowed the race down to 3 very similar logical contenders and a few non contenders that as a group I think have a similar chance of pulling off an upset.

So 3-1 is more or less my breakeven point on the contenders.

As I start picking through the information on the contenders and I notice:

1. Beyer has one of the horses faster than CJ and Thorograph has different figures than both of those guys.

2. One of the horses is coming off a track that may or may not have been biased but there haven't been any run backs yet for me to make a solid determination yet.

3. There are 2 need the lead speed horses in the race and the rest are closers. One of the speeds is coming off the turf. So I'm not sure if he'll show speed on dirt or not. If not, the remaining speed horse will have a huge tactical advantage. If he does, they could hook up.

4. One of the horses is coming out of race where the first 2 out disappointed and I know it was a tough day to make figures. So I am beginning to question the figure.

With that much confusion on the relative merits of each horse, I'm not taking 4-1 on any of those horses, let alone 7-2. I'm not getting involved unless the odds are screaming for me to play unless I can answer some of those questions well enough to be confident about the actual merits of each horse.

Now if this was rare occurrence, no big deal. But this is the kind of stuff that comes up race after race after race. My style is to avoid the confusion by passing. Others shoot for minor edges or don't even realize that there were legitimate questions. So it's quite possible I pass in situations where I should have bet. But it's also possible that most people are making marginally bad plays they don't know about.

[Light]

Rather than ROI, What I try to do is increase my bankroll by 2% every day.

This is the key to being a successful horse player. And your starting bankroll is whatever your comfort zone allows. 2% compounded will create a huge ROI.

[DeltaLover]

This is the key to being a successful horse player. And your starting bankroll is whatever your comfort zone allows. 2% compounded will create a huge ROI.

Being the key or not, growing one's bankroll by any x percentage by a specific
day, is simply wishful thinking and nothing more than that. Anyone who has some
understanding about random events should be well aware that they are associated
with very high variance, than makes such expectations completely unrealistic.

When betting horses (or participating any other form of gambling) it is of
paramount importance to decouple our approach from been result oriented. We
should be thinking in terms of expected value rather than actual short term
results. The best we can do as bettors, is to place our bets in the best
possible way, based in our opinion about the errors committed by the crowd. More
than this we have absolutely no control over the final outcome of a particular
event thus it is very realistic to expect long losing streaks, despite of our
level of competence. Although I usually refrain from using stories to support my
arguments, I am now tempted to repeat a situation that came up recently during a
conversation I had with Thask. Few months ago, I hit a cold double paying more
than $400 (based in a $2 bet!), having a large bet on it to bring down its
return from over $700 to what it finally paid. Of course this was a great boost
for my ego and my abilities as bettor.. Guess what, this feeling did not last
long, as it was followed by a very depressing losing streak of more than 25
consecutive losing bets, taking back all my profits and some more. I am sure
that all of us have similar stories to tell, but what I find interesting is the
fact that our selective memory, is doing a pretty good job, always find excuses
for our loses and near hits while exaggerating our big hits.

[traynor]

It seems that there is confusion between "paper profit" and "ROI." For bettors, ROI is (or should be) a net figure--total return, minus all investment, costs, and expenses. That is a fairly simple calculation. Paper profit is irrelevant as other than a possible indicator--it is the return on investment that is significant. No investment, no return. For bettors, ROI is equivalent to that figure presented to the IRS as income--gross wins minus all wagers and relevant costs and expenses.

The reason professional bettors find paper profit declared as ROI misleading is equally simple. If there was no wager made, there is no return--"ROI" based on paper profit from races in which no (personal) wagers were made is just juggling numbers.

Before declaring ROI as one amount or another, it might be well to indicate exactly what is meant by one's use of term. Is it an actual return in real money over some span of time (the generally accepted meaning), or is it just numbers cranked out by a software app?

[raybo]

It seems that there is confusion between "paper profit" and "ROI." For bettors, ROI is (or should be) a net figure--total return, minus all investment, costs, and expenses. That is a fairly simple calculation. Paper profit is irrelevant as other than a possible indicator--it is the return on investment that is significant. No investment, no return. For bettors, ROI is equivalent to that figure presented to the IRS as income--gross wins minus all wagers and relevant costs and expenses.

The reason professional bettors find paper profit declared as ROI misleading is equally simple. If there was no wager made, there is no return--"ROI" based on paper profit from races in which no (personal) wagers were made is just juggling numbers.

Before declaring ROI as one amount or another, it might be well to indicate exactly what is meant by one's use of term. Is it an actual return in real money over some span of time (the generally accepted meaning), or is it just numbers cranked out by a software app?

If all racing costs come out of your "available bankroll", as it should be, then ROI is also affected by all those costs, if you are betting a percentage of available bankroll, and your ROI is legit, plain and simple. If someone pays some of the racing costs with non-bankroll funds, then their ROI is not perfectly legit. I for one, take all racing costs from my available bankroll, except internet cost which I would be paying anyway, whether I played the races or not. I do not travel to the track to play so I have none of those costs to include, only my time, my wagers, and my data costs (which are extremely low).

[green80]

So, you're saying that everyone here loses 50% of their wager costs long term? Man, why would anyone play the horses, so they can lose money? There are many more enjoyable ways of losing money, wouldn't you agree? If I lost any percentage long term I would not be playing. I also wouldn't be playing if I was only showing a couple of percent profit, it just wouldn't be worth the time and effort.

No, I mean nobody has a 50% profit, maybe I worded that wrong. I think that if someone makes a roi (profit) of 5-7% they are doing better than most. Anything over 10% they would get rich quick. The big operations are happy to break even on their bets and profit from the rebates.

[nat1223]

how often do you play a day and how many days do you play a week? how many bets do you make in a year? About 4.8 per hour of live play.




What is "PNL?"




Delta, these have been my feelings exactly for the last 20 years; ever since the Beyer number went into the DRF.

The mentoring I received from someone who was doing exactly that - what what was inconprehensible to me - changed all of that.

[traynor]

No, I mean nobody has a 50% profit, maybe I worded that wrong. I think that if someone makes a roi (profit) of 5-7% they are doing better than most. Anything over 10% they would get rich quick. The big operations are happy to break even on their bets and profit from the rebates.

I would caution against confusing ROI with an advantage. The latter suggests exponential growth--like a 5% (or whatever) casino advantage that is active on each and every transaction. The more you play, the more you lose.

ROI--used accurately--is the return on a specific amount invested during a specific time frame. A totally different concept.

If one had a 5% advantage wagering on horse races, it would take a relatively short period of time to take all the marbles and send everyone else home crying for mommy. A 5% (real world) ROI could be someone betting $1000 a month for a year, and ending the year with a net profit of $600. Calculated on a monthly basis, that means betting $1000 to earn a profit of $50. That person could--in all honesty--declare that he or she had an ROI of 105% (1.05). BTW, the 5% would be more accurately labeled POI (Profit on Investment). I think you intended the 50% figure to represent profit (and an ROI of 150% or 1.5).

The reason most heavy bettors ignore paper ROI figures, however well-intentioned the people claiming them, is because significant amounts dumped into the mutuel pools tend to make those paper ROIs evaporate at an alarming rate. The only "real" ROI figures are just that--the actual return on actual investments over a specific span of time (preferably unrelated to the last big win as a marker point).

[raybo]

No, I mean nobody has a 50% profit, maybe I worded that wrong. I think that if someone makes a roi (profit) of 5-7% they are doing better than most. Anything over 10% they would get rich quick. The big operations are happy to break even on their bets and profit from the rebates.

A 50% profit means very little, I had many times that after the 3rd month I wagered full time. What means something is the return on the total amount risked, because now you're also risking the profits you make along the way, profits that you could have pocketed instead of betting them. Profit could mean profit over and above your initial investment (POI). Say I deposited $400 in a betting account and a month later I had $600 in that account, without making any other deposits, that's a 50% profit, regardless of how much I wagered during that month time period. POI, comparing initial investment to current balance is like investing a fixed amount of money in a stock and then looking at the value of that stock at a later point in time, having made no further investments. In horse racing, POI means very little unless you never lose your initial investment (your beginning total available bankroll).

ROI as it is meant to be used, means not only your initial investment, but also any subsequent investments, and/or re-investments of your individual profits along the way. So, total amounts paid back to you divided by the total amounts wagered, in that time period. If you invest $100 initially and then proceed to wager $1000, without losing that initial $100 investment, and you obtain payouts from those wagers totaling $1500, then your ROI for that time period would be $1500/$1000 = 1.50 (1 1/2 times the total amount you risked, while your POI would be ($1500-$100)/$100 = 14.00 (1400% profit, or 14 times your initial investment). Big difference! Hope I got all that right! :lol:

[Dave Schwartz]

how often do you play a day and how many days do you play a week? how many bets do you make in a year?

As I have stated, the live play part is new to me. I have been testing, testing, testing, with occasional live plays over the last 4-5 months.

I began betting according to my plan about a month ago, but have only played 10 days, averaging (as I said) about 4-5 bets per hour. I won all 10 days, but most of the days were between 1.5 and 3 hours. Yesterday I played 4.5 hours.

I find being "in the pit," as I call it, very fatiguing. Thus I am trying to work up to it.

My expectation is that I will be able to handle about 25 hours per week within a couple of weeks and, perhaps, will try some multi-sessions per day. Ultimately, I would like to get to the point of around 300 wagers per week.

I hope that answers your question.

If you are asking because you would like to know how many races are in my sample, I have made 111 wagers since beginning play, with a $2.38 $Net (+19% ROI). My testing period is slightly over 800 bets. The $Net is almost identical in both samples.

[Robert Fischer]

This is the key to being a successful horse player. And your starting bankroll is whatever your comfort zone allows. 2% compounded will create a huge ROI.

Being the key or not, growing one's bankroll by any x percentage by a specific
day, is simply wishful thinking and nothing more than that. Anyone who has some
understanding about random events should be well aware that they are associated
with very high variance, than makes such expectations completely unrealistic.

When betting horses (or participating any other form of gambling) it is of
paramount importance to decouple our approach from been result oriented. We
should be thinking in terms of expected value rather than actual short term
results. The best we can do as bettors, is to place our bets in the best
possible way, based in our opinion about the errors committed by the crowd.

Delta, I actually happen to agree with what you are saying here.

The 2% bankroll increase works for me. However, I still wouldn't say that this contradicts your point.

Important to mention - I never change the frequency or size of a wager in order to reach a goal.

I do have a simple chart set up on excel. Each 2% advancement in bankroll is a "day". I can advance or lose "days". It helps me to focus on short-term progress, and provides a visual countdown of "days" to reach long term goals.

[Light]

Few months ago, I hit a cold double paying more
than $400 (based in a $2 bet!), having a large bet on it to bring down its
return from over $700 to what it finally paid. Of course this was a great boost
for my ego and my abilities as bettor.. Guess what, this feeling did not last
long, as it was followed by a very depressing losing streak of more than 25
consecutive losing bets, taking back all my profits and some more.

I know a guy who literally went though $1 million in 5years. A good friend. Along the way he won 10 grand here and 20 grand there. In the end he lost it all plus over a million dollars.Very unimpressive.

Because like most players, he has no discipline and couldn't be bothered till this day. So your example has no validity to what I said. It's a proven road to ruin.

If you grow your bankroll by 2% (as I stated) in a disciplined manner, then your focus is on that and not on hitting a huge DD or any other exotic.This means you pass a lot of races and you don't play high risk exotic bets that will give you long losing streaks you can't handle.

[Light]

A Clarification. RF's approach of increasing your bankroll by 2% per day is not exactly my approach but it is the closest to mine and is the most sensible one presented in this thread IMO.

[TheEdge07]

Keep it simple like Crist says money wagered......and money in my pocket when day has ended.

[traynor]

ROI is simple. It has nothing to do with shuffling money form one pocket to another, or calling one chunk "my money" and another chunk "the track's money."

For a single-period review divide the return (net profit) by the resources that were committed (investment):

return on investment (%) = (Net profit / Investment) × 100

or

return on investment = gain from investment/ cost of investment
http://en.wikipedia.org/wiki/Return_on_investment


"Original bankroll" or "betting fund" is not a factor. The cost of investment is the total of all wagers, plus associated costs and expenses. The ROI is the total of all winning wagers--the gain from investment--divided by the cost of investment.

As Lawrence Revere stated so eloquenlty, "It doesn't matter who wins or who loses. The important thing is who ends up with the money."

[raybo]

I use a percentage of bank for bet sizing, that percentage varies by track depending on my hit rate at that track. I also have a target profit percentage (percentage of bankroll) that triggers my bet increases. So, my bet size is not strictly based on percentage of current bankroll, the target profit percentage must be reached before a bet increase is implemented. However, if I reach that target percentage via a single large payout, I do not increase my bet size. I have found that increasing my bet after a large payout that puts me over the target percentage, leads to a much more rapid loss of that last profit when losing streaks occur. I also have stop/loss rules that notify me when a decrease in bet size is warranted, in order to limit the loss of previously gained profits. In this way, all those previous profits must be lost before my starting bankroll is again at risk.

[green80]

I use a percentage of bank for bet sizing, that percentage varies by track depending on my hit rate at that track. I also have a target profit percentage (percentage of bankroll) that triggers my bet increases. So, my bet size is not strictly based on percentage of current bankroll, the target profit percentage must be reached before a bet increase is implemented. However, if I reach that target percentage via a single large payout, I do not increase my bet size. I have found that increasing my bet after a large payout that puts me over the target percentage, leads to a much more rapid loss of that last profit when losing streaks occur. I also have stop/loss rules that notify me when a decrease in bet size is warranted, in order to limit the loss of previously gained profits. In this way, all those previous profits must be lost before my starting bankroll is again at risk.

I also use a percentage of bank for bet sizing. I don't have a formula for adjusting this percentage for hit rate. Any suggestions?

[raybo]

I also use a percentage of bank for bet sizing. I don't have a formula for adjusting this percentage for hit rate. Any suggestions?

Yes, lots of testing!

No seriously, the higher the hit rate, the shorter the expected losing streaks, allowing a larger percentage of bank as the starting bet size.

[Dave Schwartz]

My expectation is that I will be able to handle about 25 hours per week within a couple of weeks and, perhaps, will try some multi-sessions per day. Ultimately, I would like to get to the point of around 300 wagers per week.

Oops. I just realized I had a typo in my previous message.

I meant to say "200" wagers per week.

[whodoyoulike]

... I would like to get to the point of around 300 wagers per week.


Can I ask you several questions? How do you select your races? Do you visually review which race you will be handicapping? I realize your process is mostly automated. When you say 300 wagers, are you betting on 300 races? I'm curious because I usually find less than 10 bettable races per week.

Thanks,

[Dave Schwartz]

First, that was a typo. Should have been 200.

I handicap every race that does not have a FTS and I have a reasonable minimum of races similar to in my database in terms of pace pressure at this track-surface-distance-age.

Basically, about 90% of the races without a FTS.

Other than the above, I do absolutely no "looking at the race or conditions." Fast track, off track, does not matter.

I handicapped each race using a set of rigid rules (which takes about 3 minutes). Using my rules, I wind up with some contenders - usually about 3.5 per race - and an "Order of Preference" (OOP) for putting the horses in.

Example:

Let's say that I have an 8-horse field with 4 contenders:

1
2
3
4

The OOP is:
2
4
3
1

I then begin putting horses into the bet, Dutching as I go.

Then I determine what percentage of my estimated pool is being bet, not bet or not even a contender.

The higher the percentage of the pool not being bet and not "contended" the more I wager in the race.

Simply put, if I had a race where my contenders were (in OOP):
5/2
3/1
4/1

There isn't going to be much of a bet.

On the other hand, if I had a race that looked like (in OOP):
2/1
3/1
8/1
12/1

The formula would likely call for a small wager on 2/1 and 3/1 or 2/1 and 8/1.

If the OOP was
3/1
8/1
12/1
2/1

I would have a larger wager because the 2/1 was not going to be wagered.
If the OOP was
8/1
12/1
3/1
2/1

I would have a still larger wager.

If the OOP was:
4/1
12/1
8/1
3/1


The wager would get still larger.

About 75% of the races I handicap are betable.

[whodoyoulike]

What does FTS mean? With 4 bets per race you're still finding about 40 - 50 races per week? Do you bet Mdn & mdn claiming? If you do, how do you treat 1st time starters? Which is why I don't bet those races and there seems to be a lot races under these conditions. I really appreciate your replies and would like to thank you in advance.

[Dave Schwartz]

FTS is first time starter.

I did not say I was BETTING 4 per race. Last time I noticed, I was averaging about 2.2.

I do all kinds of races. Except jumps. LOL

Oh, and nothing shorter than 4.5f.

[Dave S]

Dave:

I thought that you were using your adjusted morning line odds to bet into and not the actual tote odds- because of the problem with late odds changes.

[Dave Schwartz]

That is correct.

I use my "Object Odds" to make the bets.

[raybo]

FTS is first time starter.

I did not say I was BETTING 4 per race. Last time I noticed, I was averaging about 2.2.

I do all kinds of races. Except jumps. LOL

Oh, and nothing shorter than 4.5f.

Very similar to my win play. I too bet around 2.x (+-) horses per race, depending on the track and the method I'm using for that track. I use natural odds as a minimum odds requirement most of the time (in order to at least break even if any of my picks hit), except when playing some specific tracks, like Belmont, Keeneland, etc., where I will play my top, or my top 2 picks, only, at any odds or 1/1 odds or higher.

I don't dutch, rather I bet the same amount on each bet-able pick, (I really enjoy having the full bet size on horses that pay $60 or higher!). And, I don't vary my bet amount based on perceived value, strictly percentage of bankroll (which progresses only when I reach my next target profit percentage (a set percentage of current bankroll)) divided by the number of bets in a race. I also have a stop/loss notification built into my bet sizing app, to warn me of changes in the track environment, horses/trainer/jockey colonies or track biases, persistent changes in weather conditions/surface conditions, etc., which forces me to stop and think about reducing my bet size until the losing streak reverses itself.

I usually have 2-3 win contenders, depending on field size and the method I'm using. Sometimes I'll pass an otherwise bet-able race if one of my contenders is very low priced and looks unbeatable, but otherwise if it's a "Play" notification, I just bet the qualified contenders, without regard to any other handicapping/subjective analysis, so my win method is truly a black box method. I don't play 2 yo races as a general rule, but from time to time I will play races that have a FTS or 2 if those trainer's FTS ITM percentage is low (below 40-50%). 20%, or higher, of the field having no qualified velocities is an automatic pass for me in all of my rankings methods, and I receive a "Pass" notification from my program for those races, as well as for a couple of other pass notifications depending on the ranking method I am using at a particular track and the makeup of the field among other things. I play about 50% of the races on a card, on average, again depending on the track and the types of races carded at those tracks, and the day of the week of course.

[PICSIX]

For several years I have bought into the idea that a reasonable expectation was about 3% (before rebate).

(I think Barry Meadow's quote about that may have influenced me.)

In the last 6 months I have been mentored by a very successful player. I now believe... I almost have a hard time saying this publicly... I believe that 15-20% is possible before rebate, just as it was back in the early 90s.

Before anyone jumps on me to show proof, or accuse me of just bragging, I can only tell you that this has been my relatively recent experience over hundreds of races.

Truthfully, I am not so convinced that it has much to do with my system of handicapping or even my software. Sure, I like them both, but I have come to believe that the primary difference is my attitude.

When I say "attitude," I am referring to several key points:


Conviction that my system of handicapping is good enough to succeed without any more tinkering. (I do have a tendency to tinker with everything.)

Taking the time to paint a clear picture (of my perception) of each race.

Correctly determining which races should be strongly bet and which should be bet weakly (or not at all).

Correctly managing the money between the races (i.e. determining how much money is available to play a given race based upon current session bankroll).

Sufficient Development of courage to make the play that the system calls for despite not having lots of confidence in the outcome. (The most profitable plays rarely look much like sure things.)


Dave, someone with your skill-set should easily maintain a 15-20% return with the 200-300 plays per week.

I have left the idea of an automated system (for me) & have gone back to pen & paper handicapping. I have had a positive ROI 6 weekends straight (weekend warrior) since back to good old fashion handicapping.

This past weekend (Friday-Sunday) I had 106 win bets with a 25.47% hit rate & a 1.25 ROI or 25%.

Good luck in your pursuit! :ThmbUp:

[Dave Schwartz]

P6,

That is awesome.

I hope the time comes when I can say any double-digit ROI is "easy."

[thaskalos]

Dave, someone with your skill-set should easily maintain a 15-20% return with the 200-300 plays per week.

It has always been my opinion that high-volume play and a high ROI don't go together.

A 20% ROI coupled with making 300 plays a week means an average profit of $3,000 a week before rebates...WHILE BETTING ONLY $50 PER PLAY!!

$6,000 a week in profits if you should ever decide to increase your bets to $100 per play.

IMO...such results are the stuff that only dreams are made of...no matter who you are.

[Dave Schwartz]

Oh, sure. Destroy my dreams.

:lol: :lol:

[thaskalos]

Oh, sure. Destroy my dreams.

:lol: :lol:
I enjoy dreaming too, Dave...but, unfortunately, they don't last long...and I eventually have to wake up. :)

[raybo]

It has always been my opinion that high-volume play and a high ROI don't go together.

A 20% ROI coupled with making 300 plays a week means an average profit of $3,000 a week before rebates...WHILE BETTING ONLY $50 PER PLAY!!

$6,000 a week in profits if you should ever decide to increase your bets to $100 per play.

IMO...such results are the stuff that only dreams are made of...no matter who you are.

How many plays do you make per week? How many races do you cull those from? It's all relative Gus. If your method is automated, as many of ours are, and you look at every track running, every day, culling out the races that will not be played, and you are making multiple bets in each race, then the possibility of these kinds of returns are not unreasonable. On the shorter end of the spectrum, if you play 5 tracks per day, and you bet 50% of the races at those tracks, then you are playing about 25 races per day, or maybe 100 or so races per week, and if you are betting 3 horses per race, then you are making 300 bets per week (rebates come into play the more individual bets you make). Again, those kinds of returns are not unreasonable, if you are dutching, as Dave does. With a hit rate of 70-80%, because you are playing 3 horses per race, and the dutching insures a profit each time you hit the winner, then why is 15-20% ROI unreasonable? Then you can also add in other wager types, the horizontals and verticals, and all of a sudden you are making several more bets per day, bringing your total bets per week up well past 300, and you know, as well as I, that there is tremendous profit potential in the exotics, so that 15-20% ROI for win betting becomes 30-50% ROI, or higher, in totality.

Now, don't throw me to the wolves on those examples, I'm just putting possibles out there and have no stats to support any of that. So, take it as it is, possibilities.

I think, personally, that too many people are stuck in their own methodology, to such an extent, that they can't fathom returns of this magnitude, simply because they are only looking at a small piece of the available pie. For example, how do the syndicates manage to bet so much money in order to garner all their rebate money? Because they are looking at the whole pie, not a single piece.

No disrespect meant, but there are, really, options out there that most people are unaware of, or are unwilling to explore, or are afraid to try, etc..

[thaskalos]

How many plays do you make per week? How many races do you cull those from? It's all relative Gus. If your method is automated, as many of ours are, and you look at every track running, every day, culling out the races that will not be played, and you are making multiple bets in each race, then the possibility of these kinds of returns are not unreasonable. On the shorter end of the spectrum, if you play 5 tracks per day, and you bet 50% of the races at those tracks, then you are playing about 25 races per day, or maybe 100 or so races per week, and if you are betting 3 horses per race, then you are making 300 bets per week (rebates come into play the more individual bets you make). Again, those kinds of returns are not unreasonable, if you are dutching, as Dave does. With a hit rate of 70-80%, because you are playing 3 horses per race, and the dutching insures a profit each time you hit the winner, then why is 15-20% ROI unreasonable? Then you can also add in other wager types, the horizontals and verticals, and all of a sudden you are making several more bets per day, bringing your total bets per week up well past 300, and you know, as well as I, that there is tremendous profit potential in the exotics, so that 15-20% ROI for win betting becomes 30-50% ROI, or higher, in totality.

Now, don't throw me to the wolves on those examples, I'm just putting possibles out there and have no stats to support any of that. So, take it as it is, possibilities.

I think, personally, that too many people are stuck in their own methodology, to such an extent, that they can't fathom returns of this magnitude, simply because they are only looking at a small piece of the available pie. For example, how do the syndicates manage to bet so much money in order to garner all their rebate money? Because they are looking at the whole pie, not a single piece.

No disrespect meant, but there are, really, options out there that most people are unaware of, or are unwilling to explore, or are afraid to try, etc..

There is no disrespect meant from my side either...I am only stating an opinion as clearly and respectfully as I can state it.

Dave has explained his method of play...and I doubt that it involves exotic wagering. There is nothing wrong with betting on two of three horses to win in the same race...but overdoing this does have a negative effect on the overall ROI...in my opinion.

Look...I don't want any misunderstandings here. I don't have a jealous bone in my body...and it pleases me enormously to see those around me having great success in this game. And I have never maintained that something was impossible to do...just because I myself couldn't do it. But we live in a pretty big world...and if no one in the WORLD has been able to do something...can I be blamed for thinking that this something just might be impossible to do?

Maintaining a 20% ROI while making 300 plays a week means that we can take a relatively small bankroll of $1,000 and -- while making conservative win bets of $10 per play -- expect to double this bankroll after an average of two weeks of play. Of course, we cannot expect that the bankroll will double every two weeks...but it can just as easily double even sooner than that. That's how "averages" work. Now...do I have to tell you what happens to a $1,000 bankroll...when it doubles after every couple of weeks?

This type of thinking looks feasable in theory...but many have put the theory into practice...and the final results have always disappointed in the end.

I don't relish destroying people's dreams. I do a fair amount of dreaming myself.

But, ultimately, it isn't the dreaming that sustains us...

[baconswitchfarm]

I will also agree. I am not saying it can't be done. To have a 1.20 roi with significant play seems ambitious. I don't know anyone to ever come close to those numbers. To say it is a reasonable number is nonsense. All the pro players I know,myself included, have never been close. If you think being three to four hundred percent better than the top one percent of players is reasonable, well ok then. It would be like saying I am working on a new swing and should bat .700 in the majors next year. Nice in theory. If you get it worked out you will be living next door to bill gates when you retire. Would love to see you prove us wrong.

[raybo]

There is no disrespect meant from my side either...I am only stating an opinion as clearly and respectfully as I can state it.

Dave has explained his method of play...and I doubt that it involves exotic wagering. There is nothing wrong with betting on two of three horses to win in the same race...but overdoing this does have a negative effect on the overall ROI...in my opinion.

Look...I don't want any misunderstandings here. I don't have a jealous bone in my body...and it pleases me enormously to see those around me having great success in this game. And I have never maintained that something was impossible to do...just because I myself couldn't do it. But we live in a pretty big world...and if no one in the WORLD has been able to do something...can I be blamed for thinking that this something just might be impossible to do?

Maintaining a 20% ROI while making 300 plays a week means that we can take a relatively small bankroll of $1,000 and -- while making conservative win bets of $10 per play -- expect to double this bankroll after an average of two weeks of play. Of course, we cannot expect that the bankroll will double every two weeks...but it can just as easily double even sooner than that. That's how "averages" work. Now...do I have to tell you what happens to a $1,000 bankroll...when it doubles after every couple of weeks?

This type of thinking looks feasable in theory...but many have put the theory into practice...and the final results have always disappointed in the end.

I don't relish destroying people's dreams. I do a fair amount of dreaming myself.

But, ultimately, it isn't the dreaming that sustains us...

Cool! I'm not saying that Dave is doing everything I mentioned, but only putting possibles out there. And, I'm no dreamer either, I quit that stuff years ago. I'm a realist, but one with a fairly open mind to the possibilities. I have also proven to myself that much of what we hear as fact, simply isn't fact, but rather someone's, or some group's, beliefs often colored by various underlying agendas. The "fact", if it really is a fact, that there are 2% or less long term winners in this game, should lead one to think that most of what we hear from players about the game being unbeatable, and if it is beatable it's only because of rebates or extremely low ROIs, is extremely unreliable hogwash coming primarily from the losing 98%. Since there are so many losing players in this game, is it any wonder why we are surrounded by so much negativity regarding the possibility of long term success, or the degree of that success? If it is accepted that there are at least some people having long term success in the realm of 2-5%, before rebate, placing many bets per day, why would it be "impossible" for some to have success in the realm of 15-20%, before rebate, betting even more bets per day? After all, there are many tracks and races run every day, and many of them callable with a single pick, and certainly callable with multiple picks.

I, personally, have no knowledge, nor do any of us, that nobody in the world is doing what we're discussing here. I do know, for a fact, that 15-20% ROI betting fewer races, only because I play 1-3 tracks per day, on average, rather than 5 or more as in my previous post, is possible, because I am currently doing better than that, and have been for almost 2 years now, betting win only (I'm sorry I can't say it was longer than 2 years, which is no proof of it continuing to that extent in the future, but I did not have an automated win method, nor did I play win bets, before that). That does not include the superfectas, and very occasionally exactas and trifectas, due to field sizes, and supers not being unavailable, that I also play, which produces quite a bit more than the 15-20% ROIs we're talking about. Whether you believe that is up to you, and if you choose to not believe it, that's ok, it is my opinion that many people tend to believe that which is not true, just as often as they choose to disbelieve that which is true, especially in horse racing.

I think it is important that we not blindly believe what "everybody" tells us, or everything we read, and have heard and read all these many years. Nobody knows what everybody is doing, as you well know. So, there is no way we can say that "no one in the world" is doing something, because it is, literally, impossible for us to know about every one in the world, and what they are or are not doing.

Just because something sounds or appears logical, doesn't mean that it is. People do things all the time that is generally perceived to be impossible by most. I'm not saying that such things are widespread, because if they were then we would hear of it more often and our perception of what is impossible would have changed.

I'm of the opinion that almost nothing is impossible, especially in something like racing, where so many things are closely held, and yet, opinions so freely stated, as fact, by people who have absolutely no undeniable proof. This game is populated, to a large extent, by older people, whose long time experiences, and core beliefs, in racing, are significantly and stubbornly ingrained and, to a large extent, are completely close minded. These older people, generally are the people we hear from most often, so it's just as "logical" to assume that we aren't getting a complete picture of racing and it's participants, worldwide. As far as I know, there is no true authority in racing, an authority that knows all there is to know.

Possibilities abound, we either explore them, or we don't, it's our choice.

[raybo]

Again, I think there is some confusion about ROI. You seem to be equating an ROI of 20% (1.2) to an advantage of 20%, which is a whole different thing. With a 20% advantage, exponential growth would take the whole enchilada in short order. Perhaps that confusion is why so many are disappointed in their returns from wagering.

Simply stated, an overall ROI of whatever during some span of time does not equate to a whatever advantage on every wager made.

"Exponential growth occurs when the growth rate of the value of a mathematical function is proportional (http:///wiki/Proportionality_(mathematics)) to the function's current value. Exponential decay (http:///wiki/Exponential_decay) occurs in the same way when the growth rate is negative. In the case of a discrete domain (http:///wiki/Domain_of_a_function) of definition with equal intervals it is also called geometric growth or geometric decay (the function values form a geometric progression (http:///wiki/Geometric_progression)).
...
For example, with a growth rate of r = 5% = 0.05, going from any integer value of time to the next integer causes x at the second time to be 1.05 times (i.e., 5% larger than) what it was at the previous time."
http://en.wikipedia.org/wiki/Exponential_growth

I don't think Thaskalos has them confused, he is speaking about ROI, not edge/advantage.

[Dave Schwartz]

I don't think Thaskalos has them confused, he is speaking about ROI, not edge/advantage.

uh, what's the difference?

[DeltaLover]

uh, what's the difference?


I think that yes, there is clear difference between having a high ROI and having
a large edge. I am also sure that you know about that, so I think there is some
sort of misunderstanding here.

To make the difference more apparent, let's consider an extremely conservative
bettor, like the one who posted a few months ago about a spot play occurring a
few times per year, which focusing on heavy favorites it can show a pretty high
ROI, let's say in the range of 1.80. It is pretty easy to realize that another
bettor who bets hundreds of races per week, achieving a marginal ROI of let's
say 1.02, experiences a way larger edge than the first one.

[PICSIX]

I will also agree. I am not saying it can't be done. To have a 1.20 roi with significant play seems ambitious. I don't know anyone to ever come close to those numbers. To say it is a reasonable number is nonsense. All the pro players I know,myself included, have never been close. If you think being three to four hundred percent better than the top one percent of players is reasonable, well ok then. It would be like saying I am working on a new swing and should bat .700 in the majors next year. Nice in theory. If you get it worked out you will be living next door to bill gates when you retire. Would love to see you prove us wrong.

I guess I'm the top 1% of the top 1%, for the last six weeks :lol: :lol:

[PICSIX]

I enjoy dreaming too, Dave...but, unfortunately, they don't last long...and I eventually have to wake up. :)

They may say I'm a dreamer, but I'm not the only one :)

[Dave Schwartz]

I think that yes, there is clear difference between having a high ROI and having a large edge. I am also sure that you know about that, so I think there is some sort of misunderstanding here.

I really don't get it. To me, "ROI," "edge," and "advantage" are synonyms in our game.

Help me to understand, please.

[raybo]

I really don't get it. To me, "ROI," "edge," and "advantage" are synonyms in our game.

Help me to understand, please.

If you're taking about long term ROI or edge/advantage they are essentially the same. Traynor's post, I think, was about "individual" or short term edge/advantage. And, of course, he was trying to explain how that individual/short term edge would, theoretically/mathematically, translate into exponential/proportional growth or decay. Unfortunately, he sometimes has a hard time expressing himself in ways that most people can understand it.

[raybo]

I think that yes, there is clear difference between having a high ROI and having
a large edge. I am also sure that you know about that, so I think there is some
sort of misunderstanding here.

To make the difference more apparent, let's consider an extremely conservative
bettor, like the one who posted a few months ago about a spot play occurring a
few times per year, which focusing on heavy favorites it can show a pretty high
ROI, let's say in the range of 1.80. It is pretty easy to realize that another
bettor who bets hundreds of races per week, achieving a marginal ROI of let's
say 1.02, experiences a way larger edge than the first one.

Don't you mean that the 1.02 ROI player experiences a smaller edge, in many more races, not a larger edge over the whole sample? If you have a large individual race edge, in many hundreds of races, you will soon break the bank. While a player that has a large overall edge/ROI (the 1.80 ROI player), will not necessarily break the bank, because he/she is less consistent, race to race, and the individual race pools will restrict the amount he/she can wager..

[DeltaLover]

I really don't get it. To me, "ROI," "edge," and "advantage" are synonyms in our game.

Help me to understand, please.

If you're taking about long term ROI or edge/advantage they are essentially the same.



Edge should be measured in expected value and its dimension is dollars and nothing
else ( ROI for example is expressed as pure number, with no dimension).

So, it makes sense for someone to claim that his expected value of his
$1,000 bankroll is $2,500 after a specific number of events. Note that not all
of his individual positions need to be the same amount in every event, in contrary they
can vary anywhere from zero (when passing the race) to the whole bankroll (in
theory this can become a possibility although in real world it will never
happen). So, take the two bettors of our example, the conservative and the
aggressive, who happen to start the year with exactly the same bankroll and are
having exactly the same universe of races to choose their bets from. The
conservative gambler, might accumulate a very small fraction of the profit of the
aggressive even though in terms of ROI he might look superior. Indeed he is way
more inferior than the aggressive player, who manages to find many more bets.
This is exactly why he ends up winning much more despite of his lower ROI.

He is a far superior player.

[DeltaLover]

To make our case easier to understand, lets consider the following example:

There are two players, A and B and both are presented with a sequence of 100
races.

Player A bets 80 races out of 100.

He always bets $100 or $0, and either gets back $250 or loses the whole amount.

His hit rate is 50%

His expected value for every race he does not bet is: 0

His expected value for every race he bets is

0.5 * (-$100) + 0.5 * $150 = $25

So his expected value (edge) for the 100 races will be:

EV = 80 * $25 + 20 * $0 = $2,000

His total bet will be $8,000 while his total return will be 40 * $250 = $10,000

So, he ROI will be $10,000 / $8,000 = 1.25 (note that this is a PURE number)


Player B, is extremely conservative.

He bets only 5 races out of 100.

Every time he wins he gets back $400 while when he miss he loses the whole
amount.

His hit rate is 60%

His expected value for every race he does not bet is: 0

His expected value for every race he bets is

0.40 * (-$100) + 0.6 * $300 = $140

So his expected value (edge) for the 100 races will be:

EV = 5 * $140 + 95 * $0 = $700

His total bet will be $500 while his total return will be 3 * $400 = $1,200

So, his ROI will be $1,200 / $500 = 2.4 (note that this is a PURE number)


Note that, Player A has "only" 1.25 ROI, which is much less than Player's B
who has ROI of 2.4.

Despite the ROI though, Player A will win close to three times more than B
($2,000 vs $700)

[DeltaLover]

Based in this example, Player A has a $20 EV per race while Player B only $7

This means, that assuming a 9 race card, A has a cumulative EV of $180 while B
has $63.

The EV per race card is a much better performance metric than ROI.

[Dave Schwartz]

I understand "EV." That is a new term introduced to this conversation, Delta.

I do not see where I have misunderstood ROI, Advantage, or Edge.

I do like the idea that you have introduced this into the conversation, however, because, ultimately, it is what is most important.

In comparison to Growth of Bankroll, ROI (or Edge or Advantage) are simply for bragging rights.

What matters is how fast the bankroll actually grows over a period of time.

This would be my opinion. Yours may differ.


Dave

[DeltaLover]

I understand "EV." That is a new term introduced to this conversation, Delta.

I do not see where I have misunderstood ROI, Advantage, or Edge.

I do like the idea that you have introduced this into the conversation, however, because, ultimately, it is what is most important.

In comparison to Growth of Bankroll, ROI (or Edge or Advantage) are simply for bragging rights.

What matters is how fast the bankroll actually grows over a period of time.

This would be my opinion. Yours may differ.


Dave



Not at all.
Of course we are in agreement.
Growth per time is the what matters the most and as you also say, ROI is not a good metric of it.

[raybo]

Not at all.
Of course we are in agreement.
Growth per time is the what matters the most and as you also say, ROI is not a good metric of it.

Ok, now that we know that growth per time segment is the only way to go, how do factor in variance? No 2 time periods are going to return the same amount, so aren't you really only estimating/averaging those returns, which is exactly what happens with ROI, regarding it's chances of carrying forward into the future.

I realize that ROI in and of itself, does not measure bankroll growth, which depends on amount of money churned and hit rate, so what are we really talking about here? The OP asked about ROI, not bankroll growth.

If the thread was about bankroll growth, then the more money you bet and the higher your hit rate is, the faster, and the higher your bankroll grow and become. We all know that. However, not everyone wants to bet hundreds of thousands, or millions of dollars yearly. What about the people who just want to make steady profits while risking modest amounts of capital? I suspect there are many more "conservative/low volatility" investors/gamblers in this game than there are "high risk/high volatility" investors/gamblers. And there are probably many more successful players in the former group than in the latter. So, is this really only about bragging rights, among the very few successful players? My bank is larger than yours, so I'm better than you.

These kinds of threads really begin to grind on you the longer they go on.

If someone has "THE answer" to all this stuff, please step up and furnish the proof, in terms that everyone can understand, not some theoretical math/stats example. Never mind, because that will never happen anyway.

Bet your money and just before you die, on your death bed, announce to the world "This is how I did in racing, "long term"". But, then everyone will want you to show them the proof of what you claim you achieved. So, what's the point?

The point should be that there are good players/gamblers/investors and there are bad ones, and there are many, many more bad ones than good ones. The way you measure good versus bad doesn't really matter, because there are so few good ones anyway. If your returns are higher than your costs, then you are a good one, if not then you are a bad one. After all, aren't we all talking about 1-2%, or less, of players? Who cares how much each of those 1-2% make? They're all successful players, no matter how you measure it. One CEO is a multi-millionaire and another one is a multi-billionaire, big deal, they're both rich - who cares?

[DeltaLover]

Ok, now that we know that growth per time segment is the only way to go, how do factor in variance? No 2 time periods are going to return the same amount, so aren't you really only estimating/averaging those returns, which is exactly what happens with ROI, regarding it's chances of carrying forward into the future.


Although variance is extremely important as far as bankroll behavior is concerned, I really cannot understand what you mean by your reference to ROI. What really happens is that variance is another strong argument agaist using the sparse betting approach, since it can lead to very entensive (timewise) losing streaks.

[raybo]

What really happens is that variance is another strong argument agaist using the sparse betting approach, since it can lead to very entensive (timewise) losing streaks.

This depends on your hit rate at specific tracks. That is why each copy of my program contains data for a single track. Everything I do is track specific, hit rate, minimum odds requirements, ROI, % of bankroll bet, target profit % for bet size increases, number of horses bet on, etc.. But basically, the higher the hit rate the shorter the losing streaks, which allows for making fewer, but larger, bets, producing less variance. And, of course, vice versa.

If I'm playing 3 tracks simultaneously or overlapping, then I have 3 copies of my program running at the same time. I think this is where many go wrong, they lump all their tracks together into the same method, what works at 1 track doesn't necessarily work at other tracks, so the hit rate varies, as does the average win payouts, and of course the ROI, and the overall variance.

One could go further than track specific, differentiating between surface, class, distance, etc., but if you're using a database in your method, as I do, then the sample sizes for each of those differentiations becomes too small very quickly.

[DeltaLover]

Obviously you are correct about the hit rate.

You seem to be convinced that you need to adjust your handicapping methods based
on the track. At this point, I do not want to express my opinion on this, but I
would like to ask you, if you have any methodology to prove that indeed the
track makes a difference in handicapping. I am asking because I have done a lot
of work on this subject and took me a while to form an opinion about it.

So, the question is, what really makes you to believe on the track importance as
far as horse betting is going (please notice that betting and handicapping
although related, they are still distinct concepts).

[lansdale]

This depends on your hit rate at specific tracks. That is why each copy of my program contains data for a single track. Everything I do is track specific, hit rate, minimum odds requirements, ROI, % of bankroll bet, target profit % for bet size increases, number of horses bet on, etc.. But basically, the higher the hit rate the shorter the losing streaks, which allows for making fewer, but larger, bets, producing less variance. And, of course, vice versa.

If I'm playing 3 tracks simultaneously or overlapping, then I have 3 copies of my program running at the same time. I think this is where many go wrong, they lump all their tracks together into the same method, what works at 1 track doesn't necessarily work at other tracks, so the hit rate varies, as does the average win payouts, and of course the ROI, and the overall variance.

One could go further than track specific, differentiating between surface, class, distance, etc., but if you're using a database in your method, as I do, then the sample sizes for each of those differentiations becomes too small very quickly.

The one thing that Benter and Raybo have in common, and which I'm fairly certain is the prime factor in the latter's success, is they both created track-specific models. In his famous article, Benter strongly emphasizes that his exhaustive model applied only to the limited (1200) horse population and betting characteristics of the human population of the 3-track Hong Kong circuit. The mathematical tool he used for this analysis, mononomial logit modeling is most commonly used to analyze demographics and election results. Analysis of the betting behavior of the crowd is far more important than analysis of the racing behavior of the horses.

This is really not news. I think everyone here probably knows someone who has success playing just one track or does this themselves. But it's something that many ignore to their detriment.

Despite his success, Raybo seems to me the biggest mystery on this site. Although his years of handicapping research have given him the power to transform himself into a multimillionaire, he seems to regard his handicapping as a sort of annuity. Different strokes, I guess.

Cheers,

lansdale

[Dave Schwartz]

Delta
You seem to be convinced that you need to adjust your handicapping methods based on the track. At this point, I do not want to express my opinion on this, but I would like to ask you, if you have any methodology to prove that indeed the track makes a difference in handicapping. I am asking because I have done a lot of work on this subject and took me a while to form an opinion about it.

Lansdale:
The one thing that Benter and Raybo have in common, and which I'm fairly certain is the prime factor in the latter's success, is they both created track-specific models.

I've got to agree with Lansdale here. IMHO, a global model with local models mixed in is a good one.

Example would be that the local pace profile or post position profile may carry the same weight in the global model but the values per track are different.

[raybo]

Obviously you are correct about the hit rate.

You seem to be convinced that you need to adjust your handicapping methods based
on the track. At this point, I do not want to express my opinion on this, but I
would like to ask you, if you have any methodology to prove that indeed the
track makes a difference in handicapping. I am asking because I have done a lot
of work on this subject and took me a while to form an opinion about it.

So, the question is, what really makes you to believe on the track importance as
far as horse betting is going (please notice that betting and handicapping
although related, they are still distinct concepts).

To answer both your questions, handicapping and betting, I have a track testing mode in my program that allows me to analyze each track in, at present, 11 different handicapping methods, displaying all the races played by each method, the horses selected by each method as contenders, the hit rates for each contender, the win, place, and show payouts for each contender, the ROI for each contender, the high payout, low payout, average payout, and median payout for each contender, total amount bet and paid for each contender. I also have all that data broken down into minimum odds requirements from .01/1 through 11/1 and higher. All this data allows me to analyze which method(s) produce best, in all of those categories and odds requirements, for each specific track.

From all that data, I can state, without a shadow of a doubt, in my mind, that tracks do not play the same and require different approaches and different odds requirements, in order to play them more than marginally successfully.

All that being said, I have no doubt that "some" are able to lump them all together and eek out a small edge, regarding their totality, however, someone would have to prove to me that they have more than a very small edge doing it that way, because I simply do not see, in my testing and my experience, any evidence that substantial edge can be achieved using such a method. I suspect that many "modelers" will also agree with me on this, as many/most of their models are probably track specific as well.

Like most here, I tried for years to find a method, or a few methods, that produced consistent profits, long term, at multiple tracks, but was unable to find any. Note that I'm not saying that it can't be done, but I would be very surprised, and very interested, in seeing the documentation for such a method.

[raybo]

The one thing that Benter and Raybo have in common, and which I'm fairly certain is the prime factor in the latter's success, is they both created track-specific models. In his famous article, Benter strongly emphasizes that his exhaustive model applied only to the limited (1200) horse population and betting characteristics of the human population of the 3-track Hong Kong circuit. The mathematical tool he used for this analysis, mononomial logit modeling is most commonly used to analyze demographics and election results. Analysis of the betting behavior of the crowd is far more important than analysis of the racing behavior of the horses.

This is really not news. I think everyone here probably knows someone who has success playing just one track or does this themselves. But it's something that many ignore to their detriment.

Despite his success, Raybo seems to me the biggest mystery on this site. Although his years of handicapping research have given him the power to transform himself into a multimillionaire, he seems to regard his handicapping as a sort of annuity. Different strokes, I guess.

Cheers,

lansdale

Please don't compare me, in any way, with Benter, but I am quite certain that in order to achieve the monetary success he has, he indeed employed many track specific models, and probably broke them down into even more factor specific segments.

I failed to mention that all tracks are not played by the same bettors, which will have a dramatic affect on the way and the amount one approaches the wagering aspect, thus my emphasis on the payouts and odds requirements.

I also must point out that there are tracks at which I cannot produce a profit, with my methods. Normally the reason for that is that those tracks have very low average win payouts, long term. Consequently, I concentrate the bulk of my play at tracks that have above average win payouts.

Regarding your statement about my possessing "the power to transform himself into a multimillionaire", my experience does not support that theory. I believe that the only way to produce "millions" is the polar opposite of what I am comfortable doing. IMO, one would have to be a Bill Benter, or similar, and possess the resources in technology, personnel, and infrastructure, to say nothing of the start-up capital such a venture would require, in order to achieve those levels of profits.

Heck I can't even create a decent odds line, for God's sake!

[raybo]

To add another opinion, I personally believe that through the dilution of different results for different tracks, one is also diluting the degree of success one has at individual tracks, obscuring those data points at which one is substantially profitable. Only by separating the results, at least by track, can one see where one should be concentrating his/her efforts, and to what degree.

[traynor]

I really don't get it. To me, "ROI," "edge," and "advantage" are synonyms in our game.

Help me to understand, please.

I think this may be one of those cherished areas for which an explanation is actively avoided because it would tend to diminish the high hopes and expectations that a lack of understanding allows to exist. I have been hoping that DL or TM or some other stat/math whiz would explain it.

I think they may have the same reservations I do about such explanations. No one wants to be the one to tell the kiddies there is no Santa Claus. Especially so close to Christmas.

The basic misunderstanding of the concepts involved is a direct cause of many losses by those who believe them to be synonyms, and the direct cause of the ability of "whales" to earn seven figure incomes by using apps developed by those who know the difference.

[traynor]

http://investor.gov/tools/calculators/compound-interest-calculator

Use this to calculate your "real" edge or advantage. Nothing added other than the compounding of interest, with time equated to transactions--wagers. This is a best case scenario, in which there is zero risk of loss, and the base amount grows at the rate specified--the assumed (or presumed) "edge" or "advanatge."

To really understand what is going on, solve the equation for compound interest for the unknown of "rate." That means plug in the values for starting amount and end amount (after a 20% or whatever ROI), and the number of transactions (wagers) substituted for "t" (time). When all the values except rate of interest are on one side of the equation, it will be clear that the "edge" (if it actually exists in the real world, rather than being an illusion created by anomalies in a small sample that are unlikely to repeat when applied to a larger sample) is nowhere even close to that believed to exist.

[raybo]

http://investor.gov/tools/calculators/compound-interest-calculator

Use this to calculate your "real" edge or advantage. Nothing added other than the compounding of interest, with time equated to transactions--wagers. This is a best case scenario, in which there is zero risk of loss, and the base amount grows at the rate specified--the assumed (or presumed) "edge" or "advanatge."

To really understand what is going on, solve the equation for compound interest for the unknown of "rate." That means plug in the values for starting amount and end amount (after a 20% or whatever ROI), and the number of transactions (wagers) substituted for "t" (time). When all the values except rate of interest are on one side of the equation, it will be clear that the "edge" (if it actually exists in the real world, rather than being an illusion created by anomalies in a small sample that are unlikely to repeat when applied to a larger sample) is nowhere even close to that believed to exist.

Traynor, why do you keep harping on this "edge" thing? I think most here already know there is a drastic difference between "individual race edge" and "total/long term edge". The fact that nobody truly knows what their individual race edge is, because individual races are unique events and will not perform exactly the same in all instances, would make your point mute. Sure, you can calculate an "individual race edge" for use in projection of "potential" progression of bankroll growth or decline, but beyond that, individual race edge does not exist, because none of us knows what our edge is in individual, unique races.

So, why not just tell everyone here that using ROI, as a measurement tool, is worthless, and only people dumber than you would ever do such a ridiculous thing, and be done with it? Geez! We all know that you're the smartest person on the planet, and that none of us can possibly carry our "ROI" into the future, and the only way any of us will ever prove that we have a log term positive ROI is when we die and the auditors meticulously go through every wager we ever made in our whole life, and compare all those costs plus all our related racing costs, with the amount we were paid for our wagers?

Sorry for the rant people, but this guy really gets under my skin sometimes, with his constant suppositions, and declarations, that all the players claiming profitable play here don't really know anything about being profitable, and couldn't possibly continue being profitable in the future, because we don't think the way he does, and we couldn't possibly have a profitable method, because only he, and his formally educated cronies, have the only correct methods, all scientifically proven and the rest of that crap.

This is horse racing, you either make more, long term, than you spend, or you don't. Who cares what your individual race edge is?

Mike, if I've stepped out of line here, just delete it, and I apologize in advance.

[sjk]

Having read these sorts of threads for years it seems to me that people who are doing well are sympathetic to the thought that others on the site are doing well.

When someone repeatedly harps on how that which others must be doing must not be doing well I view that as a reflection of his own experience.

[traynor]

Denial is more than just a river in Egypt.

[raybo]

Denial is more than just a river in Egypt.

And people who holler the most about "denial" usually suffered from it in the past, and possibly still do in one way or another. Look in the mirror, with both eyes open, there may be things there that you're not particularly proud of, and might want to change, but in your case you'd probably just see only what you want to see anyway. My betting account is not subject to denial.

[LottaKash]

Sorry for the rant people, but this guy really gets under my skin sometimes, with his constant suppositions, and declarations, that all the players claiming profitable play here don't really know anything about being profitable, and couldn't possibly continue being profitable in the future, because we don't think the way he does, and we couldn't possibly have a profitable method, because only he, and his formally educated cronies, have the only correct methods, all scientifically proven and the rest of that crap.

.

Ah, variety is the "spice of life"....I love the variety here...:jump:

[DeltaLover]

So, why not just tell everyone here that using ROI, as a measurement tool, is worthless

Did you read my example a few postings above? What do you have to say about it?

[whodoyoulike]

Having read these sorts of threads for years it seems to me that people who are doing well are sympathetic to the thought that others on the site are doing well.

When someone repeatedly harps on how that which others must be doing must not be doing well I view that as a reflection of his own experience.

Have these posters who denigrate others when the others have shown their methodologies actually work for them, are they able to show that they are capable of doing well themselves? (I know that was a long sentence, I kind of got lost writing it.) Raybo, I respect your candid responses. Don't be discouraged by some posters who appear to be blowing smoke up your ass because that is the time to fart and give them a little blow back.

[raybo]

Did you read my example a few postings above? What do you have to say about it?

I totally agree with it, player A will win more money than player B, without a doubt. Of course the formula where you end up multiplying by $0 should return $0, not what you have there, but I understand what you meant.

You, evidently, and others here also, are of the opinion that the only measure of success as a player, is that they make huge amounts of money. At least for me, that is definitely not the case. I know what I do works well, and I make good supplemental income, and that's good enough for me. I know I will probably not become wealthy from racing, in whatever time I have left on this planet, unless I get lucky and hit some very big supers, and that's ok, my goal is to remain in the black, at as close to present levels as possible, or higher. If I ever drop below 20% ROI, at any track, I will cease betting at that track, that's the bottom line. I know that there are many tracks available to me, so the ones that decline are replaced by others that are performing better.

Could I optimize my profits better? Yes, of course, but that would require of me more than I am willing to do, or comfortable in doing. I don't want to be a Benter, and I don't want to play every track running, and i don't want to bust my butt churning thousands and thousands into the pools so I can take 2 or 3% net profit at the end of the year, or 5 years, or 20 years from rebates, or otherwise, and I don't want to risk hundreds of dollars on a single bet. I want steady cash flow, now, and in the near future, to support my superfecta play, where the real profit making potential is at, that's it for me.

[DeltaLover]

You, evidently, and others here also, are of the opinion that the only measure of success as a player, is that they make huge amounts of money.

I do not necessary imply that the bettor should make huge amounts of money.

Something like this is impossible due to the relative small size of the pools.

Besides this, I still believe that we need some realistic chance to make at least enough to guarantee a standard of life, meaning that there is an objective minimum that has to be met.

If I am going to dedicate more time on horse racing than I do for my regular job, have expenses for buying data and going through a continuous cycle of R&D, I should at least make as much I am making from a full time 9-5 job.

I am certainly not satisfied by showing a modest profit, even if this is associated with an extremely large ROI. I am not saying that it is wrong to view the game as an intellectual challenge and try to beat it from a clearly academic prospective, what I say is that such an approach, simply does not worth it.

[sjk]

Denial is more than just a river in Egypt.


You don't seem to understand that people do not have to conjecture as to whether they are winning or not. Either your AWD has sent you a lot of money over the years or they have not.

[raybo]

I do not necessary imply that the bettor should make huge amounts of money.

Something like this is impossible due to the relative small size of the pools.

Besides this, I still believe that we need some realistic chance to make at least enough to guarantee a standard of life, meaning that there is an objective minimum that has to be met.

If I am going to dedicate more time on horse racing than I do for my regular job, have expenses for buying data and going through a continuous cycle of R&D, I should at least make as much I am making from a full time 9-5 job.

I am certainly not satisfied by showing a modest profit, even if this is associated with an extremely large ROI. I am not saying that it is wrong to view the game as an intellectual challenge and try to beat it from a clearly academic prospective, what I say is that such an approach, simply does not worth it.

Understand, but when you're retired on fixed income, it's not hard to double or triple that betting the ponies. That's plenty for me, I have enough of what I want, live on the lake, fish or boat when I want, at least when we're not in the middle of a drought and my channel is dried up - LOL. I worked hard all my life and now it's time to sit back and enjoy what time I have left. I've already seen enough of the world for my taste so I don't have much desire to travel, except maybe to CD, Sar, Bel, SA, etc.. I've had my toys in the past and, like anything else, you get tired of them anyway. I play a few rounds of golf, work in yard, or on the cars, work in Excel trying out new ideas or making things easier to operate, or look better, etc.. I guess what I'm saying is I've made pretty good money in my life, and from my experience, the stress and aggravation was not worth the reward in the long run. I'm happier now than when I made good money, and that's what I hope continues until they kick the dirt over me. I'm 64 years old, in good health, do what I want, when I want to do it, 3 kids who are doing very well, and make me proud, 3 grand kids and another on the way, and have a loving 31 year old wife, what else could one want? :lol:

[whodoyoulike]

I do not necessary imply that the bettor should make huge amounts of money.

Something like this is impossible due to the relative small size of the pools.

Besides this, I still believe that we need some realistic chance to make at least enough to guarantee a standard of life, meaning that there is an objective minimum that has to be met.

If I am going to dedicate more time on horse racing than I do for my regular job, have expenses for buying data and going through a continuous cycle of R&D, I should at least make as much I am making from a full time 9-5 job.

I am certainly not satisfied by showing a modest profit, even if this is associated with an extremely large ROI. I am not saying that it is wrong to view the game as an intellectual challenge and try to beat it from a clearly academic prospective, what I say is that such an approach, simply does not worth it.

If you really enjoy playing the horses, You should have multiple income sources. If you're capable of making a modest profit, a modest profit from several sources adds up. Example, when you walk into a successful fast food establishment, they always offer multiple menu choices. Every successful business has multiple income streams. BTW, if you really like your work, 8 hours a day should never be enough. If you enjoy the horses, you should also consider similar activities to complement it. Like being proficient in poker and black jack for additional income streams. At most casinos, you can do any of the three. When you have a losing streak in one (which is inevitable), you need to be able to seamlessly switch to another. What are the odds of having a losing streak in all three at the same time?

[baconswitchfarm]

[QUOTE=raybo]Understand, but when you're retired on fixed income, it's not hard to double or triple that betting the ponies. That's plenty for me, I have enough of what I want, live on the lake, fish or boat when I want, at least when we're not in the middle of a drought and my channel is dried up - LOL.
It is funny how everyone has dreams. Guys all the time tell me I am living their dream. I am busting my but gambling everyday to live on a lake so I can go fishing everyday. You sir are living my dream.

[traynor]

As a devout bettor advocate, I think they deserve better than being treated like mushrooms, intentionally or unintentionally.

I also think that many (if not most) of the "98% of bettors who lose" (or whatever the most popular number is) would benefit from understanding that the numbers they use don't work the way they have been told they work. That understanding might cause more than a few to find other methods that DO work, rather than endlessly replicating the same old same old in the hope of a different outcome.

I have little interest in whether one individual is "a winner" or "a loser." Or claims to be either. Math and statistics do not suddenly stop working because of one person's (extremely) limited experience. I refer those interested in using small models to Schlesinger's comments about the result of Anderson's three years of full-time blackjack play--regardless of how much he had won--as statistically irrelevant.

[Robert Fischer]

There's a difference between not being able to statistically prove that a small sample is significant, and claiming that a small sample can not be significant. This is not emphasized enough.

[traynor]

There's a difference between not being able to statistically prove that a small sample is significant, and claiming that a small sample can not be significant. This is not emphasized enough.

A small sample can be "significant" as long as the one using it happens to win. That result is serendipitous, and has nothing to do with the reliability of small samples (of dynamic, complex events) as a predictor.

It can be stated (honestly) "I used a small sample and won buckets of money." It can be stated just as honestly "I used a small sample and lost buckets of money." Neither claim "proves" anything about the reliability of small samples--they are no more than the individuals' highly subjective and personal experience in an extremely limited number of cases.

The belief in such foolishness may well be a big factor in why the 98% lose--they assume a cause-and-effect relationship in "analysis" of a small scattering of near randomly-placed data points. Great for hobbyists betting a couple of bucks for fun. Win some, lose some, it doesn't really matter.

Not so good if one is wagering the mortgage payment, or if one is wagering substantial amounts of money as a primary income source, or if one is developing software applications for serious bettors who are inclined to react badly when the models don't perform as expected. The kind that ignore all those tacky little caveats about "this is for entertainment purposes only, and should not be considered an inducement to gamble. Past performance does not guarantee future performance. And on and on and on CYA blah blah."

What one does personally is that person's own business. I think it is unconscionable to attempt to persuade others that one's personal perception of "reality" somehow overrides external reality. That is only true in lala land, not at the track. Math and statistics do not suddenly stop functioning because someone got lucky and won a few races.

[traynor]

A quick, easy, and well-written explanation:
http://pirate.shu.edu/~hovancjo/exp_read/tversky.htm

Excerpt:
"Even if the bias cannot be unlearned, students can learn to recognize its existence and take the necessary precautions. Since the teaching of statistics is not short on admonitions, a warning about biased statistical intuitions may not be out of place. The obvious precaution is computation. The believer in the law of small numbers has incorrect intuitions about significance level, power, and confidence intervals. Significance levels are usually computed and reported, but power and confidence limits are not. Perhaps they should be. "

[raybo]

As a devout bettor advocate, I think they deserve better than being treated like mushrooms, intentionally or unintentionally.

I also think that many (if not most) of the "98% of bettors who lose" (or whatever the most popular number is) would benefit from understanding that the numbers they use don't work the way they have been told they work. That understanding might cause more than a few to find other methods that DO work, rather than endlessly replicating the same old same old in the hope of a different outcome.

I have little interest in whether one individual is "a winner" or "a loser." Or claims to be either. Math and statistics do not suddenly stop working because of one person's (extremely) limited experience. I refer those interested in using small models to Schlesinger's comments about the result of Anderson's three years of full-time blackjack play--regardless of how much he had won--as statistically irrelevant.

Traynor,

You aren't saying anything that you haven't said many times in the past. We all know that you think that your way is the best way, and you and your little group of compatriots or partners or clients or whatever you want to call them, just go ahead and continue making your millions from racing, we'll enjoy reading about you when you guys make the Fortune 500. Meanwhile, the rest of us will just stumble around trying to figure out what is right for us, however terribly misguided that may be, you know, that may not even be making millions, or "hoping" we make millions, or trying to convince others that we are helping some little group of partners/clients that are making millions.

Your agenda is quite clear, to a few of us anyway, and since you refuse to offer any examples of your great knowledge and work, in the game of horse racing, due to "contractual obligations" or whatever the latest excuse is, other than spouting some statistical chatter pertaining to anything and everything except racing, why don't you try another tact, like being forthcoming, heck, I'm sure your buddies wouldn't mind your spreading around a little of the vast knowledge that you guys obviously all have, in racing. Heck, you guys could write a book or two and start another income steam, who knows. Maybe you can increase your "edge" a bit over what you obviously already have, by taking some hard earned money from us dummies.

Shoot man, if it's good enough, maybe even I will consider a huge database and the law a large numbers. They never worked for me before, but what the hey, maybe my luck with them will change. I might even do better than what I'm doing now, with my little bitty databases and small numbers.

Give us a little taste man, I might even throw in with you. How about it man, give us a peek, huh?

What's that you say, "you have no interest in doing that"? Tell me, just what do you have an interest in doing on this forum? Educate the masses? Show us the light? Take your rightful seat on the horse racing throne? Do you want us down on our knees begging for some of your vast wisdom? How 'bout we just go ahead and crown you now? But, you know, we'd really like to read your thesis first. By the way, you know some of us here are really dense and uneducated, so can you cut us some slack when you present your dissertation? Thanks, a million!

[Robert Fischer]

A small sample can be "significant" as long as the one using it happens to win. That result is serendipitous, and has nothing to do with the reliability of small samples (of dynamic, complex events) as a predictor.

It can be stated (honestly) "I used a small sample and won buckets of money." It can be stated just as honestly "I used a small sample and lost buckets of money." Neither claim "proves" anything about the reliability of small samples--they are no more than the individuals' highly subjective and personal experience in an extremely limited number of cases.

The belief in such foolishness may well be a big factor in why the 98% lose--they assume a cause-and-effect relationship in "analysis" of a small scattering of near randomly-placed data points. Great for hobbyists betting a couple of bucks for fun. Win some, lose some, it doesn't really matter.

Not so good if one is wagering the mortgage payment, or if one is wagering substantial amounts of money as a primary income source, or if one is developing software applications for serious bettors who are inclined to react badly when the models don't perform as expected. The kind that ignore all those tacky little caveats about "this is for entertainment purposes only, and should not be considered an inducement to gamble. Past performance does not guarantee future performance. And on and on and on CYA blah blah."

What one does personally is that person's own business. I think it is unconscionable to attempt to persuade others that one's personal perception of "reality" somehow overrides external reality. That is only true in lala land, not at the track. Math and statistics do not suddenly stop functioning because someone got lucky and won a few races.

After reading this post, and the paper (http://pirate.shu.edu/~hovancjo/exp_read/tversky.htm) that you posted, it turns out that I agree with you word for word.

You are one of my favorite posters here, in part because you deliver your top-class content in a "thought provoking" style. That thought provoking style is in part what draws other posters' ire ;) at times.

The Schlesinger reference on the other page got me. :)

It's a matter of perspective(and sometimes perspective alone is almost paradigm shifting...). In the light that your recent posts hit the subject of small numbers, I find them to be true.

My quote: There's a difference between not being able to statistically prove that a small sample is significant, and claiming that a small sample can not be significant. This is not emphasized enough.
was taken out of context by the media.


(just kidding to see who is awake lol.)

;) Lets try that again, my quote is from a different perspective.

With numbers and the ability to look at sample size, and to determine significance, there is a fallacy that goes along with exactly what I am saying in that quote.
A small sample is measured statistically, and it is then determined that it happens to be impossible to statistically prove that it is significant. Then you may hear that what is being measured in that small sample can not be significant, or a variation (a streak could not have occurred, etc..).

Much more common is the fallacy that you mention, and on which the TVERSKY/KAHNEMAN paper focused.

My fallacy is perhaps more often overlooked by intelligent people who have a strong grasp of statistics, and certainly one whose offenders are much better credentialed (than the general public who is vulnerable to common small number fallacies which you address well here: ).I think it is unconscionable to attempt to persuade others that one's personal perception of "reality" somehow overrides external reality. That is only true in lala land, not at the track. Math and statistics do not suddenly stop functioning because someone got lucky and won a few races.


I personally don't use a lot of statistics, in general, although I am a strong believer in 'profiling'. My point is actually not about that perspective at all.
My point is about being able to see reality beyond the limits of statistics. And this is certainly not a knock on statistic based, or number based, or data based players... That is not my point at all. In fact I would love to put some of the basic things that I look at in unique ways, into a program, which at the very least would save time finding plays and be accurate in finding/not-missing more plays.

There are statistics that are pretty much pure. Maybe these are measuring the flow of capital(pools). There are statistics that are "abstractions"(perhaps a performance figure). Sometimes I can outperform statistics with an abstract basis regardless of sample size.

Lets get back to small numbers and whether we can call them insignificant...
For example, a SUPERtrainer is on an ice cold streak this week...
His FTS ran a dud, his closer couldn't close, his first of the claim didn't move-Up,... his heavy fav. ran 3rd or 4th.
Statistically it means nothing.

Let me try to list some possibilities(i could not care less if I miss some, just go along for the point here)


He is about the same(or even going well), these are randomness
He actually is ice cold.
^ These are not meant to be a 50/50 probability or anything like that, merely an unweighted list.


Now lets look at application:


Wager as usual
or Pass races in which he is in
or Bet against that SuperTrainer with an alternative contender, or spread
If you Pass these races, you cover your ass. You had a "NEGATIVE" and because the sample was small you were not sure if it was real or a false flag. Unlike baseball, you can't get a called strike in horseplaying (The closest equivalent is needing to pay bills, a time constraint/responsibility most casual players don't have.). As long as a large part of your profits or churn don't rely on a certain factor, when that factor has a negative small number sample, you are OK to pass those situations.


Positive or Negative small number samples, may be investigated further. Occasionally you can gather enough info to wager against(or ON -in the case of small number "Positive" samples). These could be discovering statistically significant larger number data that the small number data lead you to. Or these could be areas where your specialized elite insight was outperforming abstract small-number data.






Finally the one that 'provokes' me is the example of a successful player whose success is shown to be statistically irrelevant.



Maybe in some cases, whether or not a person is successful is actually ambiguous, or even a person whose talent, level of insight, and wisdom would actually have a slightly higher probability of being less(or more) successful than they are, yet randomness has favored(or hindered) them in their career/year/week/day...


Yet some people are genuinely elite in their fields. Even (relatively) small numbers can not put any doubt onto their ability.

[traynor]

Positive or Negative small number samples, may be investigated further. Occasionally you can gather enough info to wager against(or ON -in the case of small number "Positive" samples). These could be discovering statistically significant larger number data that the small number data lead you to. Or these could be areas where your specialized elite insight was outperforming abstract small-number data.


One of the most useful skills to develop for a bettor is to continually ask (as business analysts are trained to do), "Does this stuff really mean what (I, we, they) think it means?"

Statistical processes have a big place in race analysis. They are definitely not the be all and end all of the process, and should not be expected to be. As you indicate above, statistics may be best used as indicators and suggestions of possible productive areas. Viewed in that light--rather than as iron-clad certainty--basic statistics and basic statistical processes are one of the most useful areas of race analysis for a prospective bettor to understand. No PhD's required--just critical thinking about what the numbers represent.

In the context of this thread, the most important thing to understand is that an ROI of 1.2 does not equate to an "advantage" or "edge" of 1.2 acting on every race. The ROI may be due exclusively to one or two anomalies in a small sample that skew the results, Nice to look at, nice to brag about, but not so good for betting on.

[Robert Fischer]

Statistical processes have a big place in race analysis. They are definitely not the be all and end all of the process, and should not be expected to be. As you indicate above, statistics may be best used as indicators and suggestions ofpossible productive areas. Viewed in that light--rather than as iron-clad certainty--basic statistics and basic statistical processes are one of the most useful areas of race analysis for a prospective bettor to understand. No PhD's required--just critical thinking about what the numbers represent.
If someone can successfully use statistics directly, I can't criticize that. That will be a better tool. I am a big fan of 'profiling'. It's a word that has taken on negative connotations. However I find it to be somewhat useful, and it can occasionally bring me to a point where I can do what I do. Most things, by definition are ordinary in classification and behavior. You are going to miss the anomalies if you blindly trudge ahead profiling, and you don't want to do that, even though they are rare. It still comes down to insight, but profiling is another short cut.


In the context of this thread, the most important thing to understand is that an ROI of 1.2 does not equate to an "advantage" or "edge" of 1.2 acting on every race. The ROI may be due exclusively to one or two anomalies in a small sample that skew the results, Nice to look at, nice to brag about, but not so good for betting on.

A lot of people online seem to have ROIs of 1.5, 1.2, etc... The first thing that comes to mind after filtering out those that I can actually picture beating the game and keeping somewhat accurate records, is how inefficient such a high ROI is. It may sound funny, but a pro or semi-pro who has been playing for a good amount of time, and is hitting those kind of ROI numbers, is going to want to start bringing those ROIs down. It's just the way the pools, and the rebates work. With the anomalies affecting the ROI, those can go both ways, but it's probably far more common online to hear of inflated ROIs than deflated.

[CincyHorseplayer]

I have one question that is universal we can all judge you by and one question I want to know just as a matter of wether we can judge you and you can denounce my formula no problem.

1)What percentage of your total bankroll do you put in play in a single day?

2)I love baseball stats and these 2 games are so comparably similar.To see how I am doing over any given period of bets I use a slash line that is just like a baseball line.Since the saber revolution I will bypass the AVG/OBP but I calculate AVG/Hit Rate by how many bets you receive back at least 5% of your total bankroll and OBP/Cash rate by what it says.After trying to use Woba or trying to weight everything like baseball's slugging percentage I came up with an accurate enough formula for a horseplayer's slugging %=total return divided by total bets + starting bank.

Call it trivial.But your slugging percentage,if it's far less than .400,you are a pretty pathetic player and I don't care how big your bankroll is and you should be especially embarrassed if you have a .300 SLG if you bet a million dollars.If it's completely about how much money you win and you think you are exempt from all other comparative values,then this thread is nothing more than a cave man comparison of my d*ck is bigger than yours and it's total BS.

[raybo]

Assuming one is not betting differing amounts based on "perceived" edge in individual races, IMO, one's record keeping should include a listing of all those races, showing the payouts for each betting interest in each race. One should be able to scan or scroll through all those races and look for possible outliers, not fairly representative of the list as a whole. After removing, or downsizing, those outliers, one should look at least at: ROI, hit rate, high price, low price, average price, median price, total cost, and total payout. This enables one to get a very good idea of how you have been doing, and how you should do in the near future. Without looking at all that stuff one really is only guessing at where one was, where one is, and where one should be in the near future. If the near future looks depressing then it's time to make adjustments in your method, or suspend play at that track, etc..

[DeltaLover]

I have one question that is universal we can all judge you by and one question I want to know just as a matter of wether we can judge you and you can denounce my formula no problem.

1)What percentage of your total bankroll do you put in play in a single day?

2)I love baseball stats and these 2 games are so comparably similar.To see how I am doing over any given period of bets I use a slash line that is just like a baseball line.Since the saber revolution I will bypass the AVG/OBP but I calculate AVG/Hit Rate by how many bets you receive back at least 5% of your total bankroll and OBP/Cash rate by what it says.After trying to use Woba or trying to weight everything like baseball's slugging percentage I came up with an accurate enough formula for a horseplayer's slugging %=total return divided by total bets + starting bank.

Call it trivial.But your slugging percentage,if it's far less than .400,you are a pretty pathetic player and I don't care how big your bankroll is and you should be especially embarrassed if you have a .300 SLG if you bet a million dollars.If it's completely about how much money you win and you think you are exempt from all other comparative values,then this thread is nothing more than a cave man comparison of my d*ck is bigger than yours and it's total BS.

Conceptually I agree about trying to put a large fraction of the bankroll in
action in a daily base. The problem is that as the BR ramps up, we very fast
reach a point where we are restricted by the pool size an have no choice but to
slow down on our bet sizing.

[DeltaLover]

After removing, or downsizing, those outliers, one should look at least at: ROI, hit rate, high price, low price, average price, median price, total cost, and total payout.

I think that the most important metrics should be PNL and max drawdown. As far as ROI and hit rate I see no value on both...

[raybo]

I suspect that successful players whose ROI is low, less than say 1.15 or so, and is not churning gobs of money at multiple tracks and/or betting interests in individual races, and not relying on that churn strictly for maximum profit, with or without rebates, would do much better if they took a very discriminating look at which tracks and which race types, surfaces, distances, etc., they are playing and adjust those as needed. If one's goal is not strictly maximum profits, then specialization, by the factors mentioned, is the name of the game, IMO. Once one has done that, bet size can be adjusted to help maximize one's profits, to the extent that the pools allow anyway.

[raybo]

I think that the most important metrics should be PNL and max drawdown. As far as ROI and hit rate I see no value on both...

Profit and loss (PNL for those who are not familiar with what he means) is a factor of ROI. Drawdown is a factor of hit rate, payouts, and bet sizes. All of which should be known to anyone using decent record keeping methods. PNL, as well as ROI, by themselves mean little unless taken in context with other factors, as is hit rate. I could have a 3.20 ROI or profit and loss ratio due to one big hit, or a couple of big hits. I could have a 75% hit rate due to a long winning streak or a couple of long winning streaks, and that would mean little without looking at other factors. All this stuff is relative to the individual factors involved, so one must include all those factors in one's record keeping, and involve all of them in one's record keeping analysis. Otherwise, one is only guessing at the viability of one's overall method.

When someone declares a ROI, PNL, hit rate, etc., unless they post all the plays involved with all the factors involved in those calculations, I am forced to assume they are taking into account all the other related factors. I state ROIs fairly often (and on my forum I sometimes post the list of plays also), but those ROIs aren't stated lightly, they are post adjustment ROIs, meaning they are viable ROIs, without regard to total profits made, as they are based on $2 flat bets, without bet size progression or regression techniques, and they are track specific. I do not state profit made because that, for the most part, is bet size dependent and number of plays/churn dependent.

[DeltaLover]

When someone declares a ROI, PNL, hit rate, etc., unless they post all the plays involved with all the factors involved in those calculations....

I do not state profit made because that, for the most part, is bet size dependent and number of plays/churn dependent.

The difference is that PNL is measured in dollars while ROI or hit rate are not.

In my opinion a PNL is MUCH better metric that anything else. I measure it assuming a standard bet size per ALL THE AVAILABLE races ( for example $0 OR $100) I can bet . Be doing so, it is easy to calculate the Expected Value per each race (whether if you are going to bet it or not), so you can make statements such: for every race that is carded in XYZ track (since you like to group by tracks) I expect to win / lose this dollar amount, assuming a $100 total bet. Of course this is not the only figure you should be concerned about as you somehow need to consider volatility as well. A good measurement of this might be maximum drawdown, although we can derive more sophisticated and accurate methods.

[raybo]

The difference is that PNL is measured in dollars while ROI or hit rate are not.

In my opinion a PNL is MUCH better metric that anything else. I measure it assuming a standard bet size per ALL THE AVAILABLE races ( for example $0 OR $100) I can bet . Be doing so, it is easy to calculate the Expected Value per each race (whether if you are going to bet it or not), so you can make statements such: for every race that is carded in XYZ track (since you like to group by tracks) I expect to win / lose this dollar amount, assuming a $100 total bet. Of course this is not the only figure you should be concerned about as you somehow need to consider volatility as well. A good measurement of this might be maximum drawdown, although we can derive more sophisticated and accurate methods.

I understand your preferences. My point is that, if your record keeping includes all necessary factors, you can derive whatever you want to know, in your case, projecting expected value. That is not what I want to know, I want to know if I should be playing a certain track or not, and what kinds of races from that track I should not play at all, and which methods and odds ranges I should be using for that track.

In short, what is important to you and me, may not be what is important to others.

[DeltaLover]

In short, what is important to you and me, may not be what is important to others.

That's the whole idea :)

[CincyHorseplayer]

How much percentage of your total bankroll do you put in play in one day?

Some of you are trying to eliminate every scoring stat because you think there are guys that bet 100 times a year.Or scold guys that bet less than a
mil.
I know where you are coming from.Traynor and Delta you guys influence me like no other,but I don't like when you abuse me because you think it's fun.

I'm not scared of your truths.My tiny samples that cover basically 1000-1600 bets per year.If I'm getting overall results that are solid.I'm not waiting to die for it to be statistically valid!Because there aren't enough samples.Yeah right.

I was cocky enough as a 24 year old to parlay a year of 229% ROI into an entire winter of 01 of life sustenance into 02.A couple winters of that.Did the same in 2004 successfully.I have bigger dreams but for a kid from nothing I'm allright.

To be continued.

[Robert Fischer]

How much percentage of your total bankroll do you put in play in one day?
hey CincyHorseplayer,
It just depends on what plays I happen to see in a day.
Usually I wager between 1/4 and 1/3 of my bankroll each day.
I haven't looked at Monday's card yet(I went to bed early Sunday night), but Monday's can be tough, even with Philly, Mountaineer, and Charles Town usually running.

Yesterday I wagered 56% of my starting bankroll. That's really on the high side (obviously 1/2 is more than 1/3 or 1/4). Funny thing is, I missed a couple plays. [EDIT wow saturday was a ridiculous 165% or 1and2/3]

10K is kind of magic number goal for me as far as bankroll is concerned.
2%(i've mentioned before that 2% blocks of progress is a one of the way I simplify short term goals) of 10K is $200 a day, which is like "real job" money. And when you wager 1/4 to 1/3 of 10K that is like >$3K a day, and now you can brag to old guys on a horseracing forum that you've hit the million dollar wagered milestone after a year of that, and they will either give congrats or call you a liar. ;) That is also getting into territory where your wagers affect prices. (which is the whole reason I initially started looking at that as a goal/plateau warning/etc). That is obviously a negative. The message is clear = perfect exotics to the level of your other wagers or stop growing...

So 1/3 is probably a close estimate and a little higher than reality.
On a specific wager however, I use my own bet-size calculations. It's pretty close to Kelly, but more conservative. The easiest way to describe is that the higher the chance of that bet winning, the more I can risk on a single bet.
Most of my wagers are around 1% of total bankroll for a wager(let's say I have a Thousand Dollars in the account or on my voucher = and I love a 2-1 shot to Win = Maybe I will bet $10 to Win on him... No more than $30(kind of steep 3% here, and would be reserved for those once a week type great picks, who I feel are big UNDERLAY/VALUE).
The most I have wagered on a single bet since starting to take money management as seriously as handicapping, is 10% of my total bankroll in one shot.

[Robert Fischer]

My tiny samples that cover basically 1000-1600 bets per year.
...

...

I was cocky enough as a 24 year old to parlay a year of 229% ROI into an entire winter of 01 of life sustenance into 02.A couple winters of that.Did the same in 2004 successfully.
...


1,600 bets x $20 = $32,000 churned
2.29ROI x $32,000 = $73,280 returned.

So if you started with $1000 cash, you made over 70 grand.
That is before you did it multiple years and parlayed it.
Also assuming you stuck with $20 wagers, and didn't move on to $50 or $100 wagers.

Very impressive.

[traynor]

How much percentage of your total bankroll do you put in play in one day?

Some of you are trying to eliminate every scoring stat because you think there are guys that bet 100 times a year.Or scold guys that bet less than a
mil.
I know where you are coming from.Traynor and Delta you guys influence me like no other,but I don't like when you abuse me because you think it's fun.

I'm not scared of your truths.My tiny samples that cover basically 1000-1600 bets per year.If I'm getting overall results that are solid.I'm not waiting to die for it to be statistically valid!Because there aren't enough samples.Yeah right.

I was cocky enough as a 24 year old to parlay a year of 229% ROI into an entire winter of 01 of life sustenance into 02.A couple winters of that.Did the same in 2004 successfully.I have bigger dreams but for a kid from nothing I'm allright.

To be continued.

??????
I have had no such intent. If you have taken something I have written as a personal attack, or derisive of your efforts, or some such, I apologize, but such was never my intent. Most of what I write is intended "generically"--for anyone who can get any value from it.

My primary purpose in posting on this thread is to point out that ROI and "edge" or "advantage" are not the same thing, and that it can be costly to believe they are. No personal attacks intended.

[DeltaLover]

Same here of course.

All of us, we just present our approach, trying to make our case without any other intention. It is natural and expected to run into disagreements, this does not mean that we try to provoke any kind of a flame war or even impose our views.

[raybo]

??????
I have had no such intent. If you have taken something I have written as a personal attack, or derisive of your efforts, or some such, I apologize, but such was never my intent. Most of what I write is intended "generically"--for anyone who can get any value from it.

My primary purpose in posting on this thread is to point out that ROI and "edge" or "advantage" are not the same thing, and that it can be costly to believe they are. No personal attacks intended.

It would have helped if you had worded your previous responses that way. Much of the "contempt" you receive here is due to the way you state things, more often than not, as fact. Whether or not that is your intent is beside the point, that is what often comes across in your statements.

[CincyHorseplayer]

Assuming one is not betting differing amounts based on "perceived" edge in individual races, IMO, one's record keeping should include a listing of all those races, showing the payouts for each betting interest in each race. One should be able to scan or scroll through all those races and look for possible outliers, not fairly representative of the list as a whole. After removing, or downsizing, those outliers, one should look at least at: ROI, hit rate, high price, low price, average price, median price, total cost, and total payout. This enables one to get a very good idea of how you have been doing, and how you should do in the near future. Without looking at all that stuff one really is only guessing at where one was, where one is, and where one should be in the near future. If the near future looks depressing then it's time to make adjustments in your method, or suspend play at that track, etc..

Ray horseplayer slugging percentage.These math whizzes can denounce it all day because it will prove they bet a ton of money in a dumb fashion and are not as good as they should be if they were indeed the upper echelon.Their money cannot buy proficiency.I've got you calculated at a .735 slugging percentage on your superfectas.If I judged you solely on hit rate,well there is no reason to go there Mr .080 hitter!But your slugging percentage so blows apart fashionable numeric blisses that only experience,not theory can teach.

[CincyHorseplayer]

Conceptually I agree about trying to put a large fraction of the bankroll in
action in a daily base. The problem is that as the BR ramps up, we very fast
reach a point where we are restricted by the pool size an have no choice but to
slow down on our bet sizing.

The reason I mentioned how much % put in play a day is to gauge where everyone is at with risk and reward.Plus the most experienced of players will have some input on what from their experience is a wise amount ideally.I say it's universal because my 2K is as valuable to me as someone else's 200k.Personally if I feel a race has an edge I can bet a single race 5-6%.I don't have a choke threshold now and I don't expect it to crop up as my bank gets larger.But I also think it's not wise to put in more than 20% on any given day.

To your post I just have to question how you distribute your money into pools.I split my bets roughly 60/40 win to exotics.I roll doubles and play the pick 4.Collectively even if I averaged $1000 per race each segment would only receive about $250.Not enough to KO prices except at maybe Beulah.I have confidence in each bet and know what their returns are.So I'm not diluting betting money for the sake of doing it.Each has it's own methodology.I certainly am not at that threshold.But are you deciding how to slice up like 4k per race?If so I can't even grasp that landscape.

Longstory short,after reading and having these conversations for months I'm trying to find the common ground and same language we can all speak.ROI obviously does not cut it.There has to be a common denominator that values each man's money he puts at risk.The profitability solution at the ceiling of this game has to be dealt with and I know that's what fascinates you.Me too.Even if I never ascend to being a million dollar bettor,I want to know what the limit is and how it works.I could be there some day.Not because I'm a starry eyed dreamer but because I know I can and do damage in this game.

[raybo]

The reason I mentioned how much % put in play a day is to gauge where everyone is at with risk and reward.Plus the most experienced of players will have some input on what from their experience is a wise amount ideally.I say it's universal because my 2K is as valuable to me as someone else's 200k.Personally if I feel a race has an edge I can bet a single race 5-6%.I don't have a choke threshold now and I don't expect it to crop up as my bank gets larger.But I also think it's not wise to put in more than 20% on any given day.

To your post I just have to question how you distribute your money into pools.I split my bets roughly 60/40 win to exotics.I roll doubles and play the pick 4.Collectively even if I averaged $1000 per race each segment would only receive about $250.Not enough to KO prices except at maybe Beulah.I have confidence in each bet and know what their returns are.So I'm not diluting betting money for the sake of doing it.Each has it's own methodology.I certainly am not at that threshold.But are you deciding how to slice up like 4k per race?If so I can't even grasp that landscape.

Longstory short,after reading and having these conversations for months I'm trying to find the common ground and same language we can all speak.ROI obviously does not cut it.There has to be a common denominator that values each man's money he puts at risk.The profitability solution at the ceiling of this game has to be dealt with and I know that's what fascinates you.Me too.Even if I never ascend to being a million dollar bettor,I want to know what the limit is and how it works.I could be there some day.Not because I'm a starry eyed dreamer but because I know I can and do damage in this game.

I don't know about others here, but the amount of bankroll I wager per day depends on the number of tracks I'm playing and the number of races I'm betting. The bet size/% bankroll per bet, depends on the track and how it tested. At some, I risk as low as 1/2%, at others I wager 3%-5%, on average, but I, like most others I'm sure, do not use a straight Kelly method for betting, and I also don't bet a flat % of bank, but rather that % of bank varies, upwards and downwards, until I remove money from my bank, decreasing the size of the bank and as a result the bet size, and I start all over.

Am I optimizing profits, no, but neither am I risking too much, nor am I overbetting the pool. I never reach a bet size that will adversely affect the odds on my horse(s), because those odds are where the value comes from.

Also, I bet multiple horses in races, not just 1, so I'm spreading the money out over more than 1 horse, often on horses that most are not betting. It is these types of horses, bet at the same amount as lower priced horses that create the larger ROIs, and that is what I am after, healthy complete meet ROIs, not huge profits. Shooting for huge profits, more times than not, sets you right in the wheelhouse for huge losses. No thank you!

[DeltaLover]

I say it's universal because my 2K is as valuable to me as someone else's 200k.

Depends of several conditions.

As an extreme example, think about a poker tournament on the bubble. In this case your 2K worth way more than 2K for a villain who's stack is 200K and obviously you need to be more careful about handling your stack. In this case, your best optimal strategy should be extremely aggressive and as tight the blind structure allows you. Note that the large stack can afford to adopt the opposite style, taking shots on situations that are prohibitive for you.

Betting horses, the same does not apply. I actually believe that the proper adjustment is a reversal of the small stack tournament strategy. A 200K bankroll needs to be treated very conservatively, since its size implies very large bets, so large to change the payouts meaning that there is very little room for a big proportional score. On the other hand, a small bankroll of 2K, leaves a lot of room for creative bets, shooting for a large or huge payout that will elevate us to the next level. Even if you get broke, this will never be a disaster, as the absolute size of this stack is so small, that it can always be recovered in a few weeks. Based on this, I completely agree about been aggressive with a small stack betting as high as 6% of it in a single bet. As a general rule, you can take more risks and been more aggressive the lower the size of your bankroll is.

As far as bet distribution per pool, I prefer the pools where I can have a better view of how the crowd is approaching them so this means that WIN - EXACTA - DOUBLE - QUINELLA are my bread and butter bets. Although I might take an occasional shot in something longer (PICK3), I will stay clear of longer gimmicks as I not only miss crowd's behavior but since I need to predetermine all my selections in advance, I also can miss changes, like scratches, weather conditions etc. More than this I try to keep my payouts below 600-2 for obvious reasons.

[CincyHorseplayer]

1,600 bets x $20 = $32,000 churned
2.29ROI x $32,000 = $73,280 returned.

So if you started with $1000 cash, you made over 70 grand.
That is before you did it multiple years and parlayed it.
Also assuming you stuck with $20 wagers, and didn't move on to $50 or $100 wagers.

Very impressive.

No,I didn't win 70 thousand dollars!!In the month of October 2001(I was 28 actually!) I won a little over $3300.00 dollars.I didn't work til the following March.My bills and expenses were about 2g's a month back then.I wish I would have taken it more seriously back then.I was always taught that it's better to go to work and make an honest living.What a crock of $hit!The worst times in my life have always been when I tried to aspire to this moral righteousness!

I might be the poster boy everybody is looking for simply because I don't bet the big wagers that produce economic anomalies like the pick 6.I'm a dirty azz grinder and make money off individual races.Have changed that too of late!

[CincyHorseplayer]

??????
I have had no such intent. If you have taken something I have written as a personal attack, or derisive of your efforts, or some such, I apologize, but such was never my intent. Most of what I write is intended "generically"--for anyone who can get any value from it.

My primary purpose in posting on this thread is to point out that ROI and "edge" or "advantage" are not the same thing, and that it can be costly to believe they are. No personal attacks intended.

You deride people with this fairly regularly.I see past it most of the time and I share my world with you because I do recognize that you have valuable insights to offer and you have shared some of the same human insights as well.I recognize the fact that you think of this game at the total top of money in play reality.As I ascend levels of play I expect no solace in how it is accomplished.Shooting the $hit with a fellow player on here?I see no need to bust everybody's balls as hard and as adamantly as you are here.I do take it personally.The work everybody puts in on here has paid us so we know it's not an illusion!Jesus I don't think I have bought any clothes or shoes since the 90"s with my own money!Getting to the elite.Being an earner at a high level.I get you on that.Maybe it's because I just turned 40 but I don't care about any of the trivial $hit I have valued up to this point.I've done it all.Sex,drugs,rocknroll,excess,just havin a pile of stuff.All I care about is having a pile of dough that buys me my out of having to deal with people who I hate.I've got my ears peeled for anything and anybody on that train.

[CincyHorseplayer]

I have 2 realities gnawing at both sides of the brain these days regarding betting.I'm betting about 3% per race and I'm comfortable with that.But races I pass and pass and pass.When the time comes along I love a race I have no problem betting more.And that's the other thing.In the past I could not always distinguish if a bet or what a great bet was.I had to instill the generic number of assigned bet amount or I would rob myself of payoffs by being right on what I deemed risky plays and wrong on what I deemed solid plays.I think every player at some point had this problem.Even still I put the lower type of play at a begrudgingly 2/3rds of a prime bet.But for the most part I know what constitutes the bigger money.Drastic difference than 5 years ago.

Point being,from what I got out of your post,the bigger bankroll I have I would feel no such necessity to even move unless I sniffed a killing.If I had a target amount to win for a certain period of time,I could be patient and wait for it.I would feel no need to scramble.That's what the point of this thread is.Being efficient in your investments.If watering down money for maximal return is the goal I understand if we are talking huge amounts.If not were talking dumb masquerading as badass because the bankroll is bigger!I'm not trying to argue.Just trying to see where everybody is at.This fascinates me.I spend half my life trying to think what men I admire think.The other half trying to conquer the world!Let you know when I get there!:cool:









Depends of several conditions.

As an extreme example, think about a poker tournament on the bubble. In this case your 2K worth way more than 2K for a villain who's stack is 200K and obviously you need to be more careful about handling your stack. In this case, your best optimal strategy should be extremely aggressive and as tight the blind structure allows you. Note that the large stack can afford to adopt the opposite style, taking shots on situations that are prohibitive for you.

Betting horses, the same does not apply. I actually believe that the proper adjustment is a reversal of the small stack tournament strategy. A 200K bankroll needs to be treated very conservatively, since its size implies very large bets, so large to change the payouts meaning that there is very little room for a big proportional score. On the other hand, a small bankroll of 2K, leaves a lot of room for creative bets, shooting for a large or huge payout that will elevate us to the next level. Even if you get broke, this will never be a disaster, as the absolute size of this stack is so small, that it can always be recovered in a few weeks. Based on this, I completely agree about been aggressive with a small stack betting as high as 6% of it in a single bet. As a general rule, you can take more risks and been more aggressive the lower the size of your bankroll is.

As far as bet distribution per pool, I prefer the pools where I can have a better view of how the crowd is approaching them so this means that WIN - EXACTA - DOUBLE - QUINELLA are my bread and butter bets. Although I might take an occasional shot in something longer (PICK3), I will stay clear of longer gimmicks as I not only miss crowd's behavior but since I need to predetermine all my selections in advance, I also can miss changes, like scratches, weather conditions etc. More than this I try to keep my payouts below 600-2 for obvious reasons.

[CincyHorseplayer]

1)Horseplayer Slugging Percentage=total return divided by total bet + starting bankroll

2)Bankroll percentage put in play per day?


Everybody is a bunch of hadas$es when it comes to busting another guy's balls.Let's cough it up ladies.

[lansdale]

Please don't compare me, in any way, with Benter, but I am quite certain that in order to achieve the monetary success he has, he indeed employed many track specific models, and probably broke them down into even more factor specific segments.

I failed to mention that all tracks are not played by the same bettors, which will have a dramatic affect on the way and the amount one approaches the wagering aspect, thus my emphasis on the payouts and odds requirements.

I also must point out that there are tracks at which I cannot produce a profit, with my methods. Normally the reason for that is that those tracks have very low average win payouts, long term. Consequently, I concentrate the bulk of my play at tracks that have above average win payouts.

Regarding your statement about my possessing "the power to transform himself into a multimillionaire", my experience does not support that theory. I believe that the only way to produce "millions" is the polar opposite of what I am comfortable doing. IMO, one would have to be a Bill Benter, or similar, and possess the resources in technology, personnel, and infrastructure, to say nothing of the start-up capital such a venture would require, in order to achieve those levels of profits.

Heck I can't even create a decent odds line, for God's sake!

Hi Raybo,

I meant this as I positive comment about an important aspect of Benter's model that is usually overlooked which has something in common with yours, as you've described it, but you seem a little defensive. I wasn't implying that you could or should want to bet on that scale.

The point of my 'multi-millionaire' comment, which is not addressed to you alone, but to *anyone* who has been playing with an edge, or whose model displays positive EV over a sustained period. For anyone who has such an edge, their bank could (or should) be increasing exponentially. Thaskalos made the same point in another thread. Or to paraphrase the great Dick Schmidt, 'If you're playing with a 30% ROI, you don't need a money management system, just a place to stack your money'.

Now, I understand from your posts here that you are quite risk averse, which, as I said, can seem mysterious, considering the obvious time and energy you've invested in handicapping, but that implies no disrespect. Actually, the opposite. And, as I said, 'different strokes'.

BTW, no need to respond to the posts of the serial narcissist who frequently pops up on this site to undermine anyone who claims to be making money gambling. As Shakespeare suggested, Iago specializes in bad advice.

Best,

lansdale

[lansdale]

Having read these sorts of threads for years it seems to me that people who are doing well are sympathetic to the thought that others on the site are doing well.

When someone repeatedly harps on how that which others must be doing must not be doing well I view that as a reflection of his own experience.

Big time understatement.

Cheers,

lansdale

[lansdale]

You don't seem to understand that people do not have to conjecture as to whether they are winning or not. Either your AWD has sent you a lot of money over the years or they have not.

Most relevant post in this thread. Very sorry you don't post here more often.

Best,

lansdale

[lansdale]

After reading this post, and the paper (http://pirate.shu.edu/~hovancjo/exp_read/tversky.htm) that you posted, it turns out that I agree with you word for word.

You are one of my favorite posters here, in part because you deliver your top-class content in a "thought provoking" style. That thought provoking style is in part what draws other posters' ire ;) at times.

The Schlesinger reference on the other page got me. :)

It's a matter of perspective(and sometimes perspective alone is almost paradigm shifting...). In the light that your recent posts hit the subject of small numbers, I find them to be true.

My quote:
was taken out of context by the media.


(just kidding to see who is awake lol.)

;) Lets try that again, my quote is from a different perspective.

With numbers and the ability to look at sample size, and to determine significance, there is a fallacy that goes along with exactly what I am saying in that quote.
A small sample is measured statistically, and it is then determined that it happens to be impossible to statistically prove that it is significant. Then you may hear that what is being measured in that small sample can not be significant, or a variation (a streak could not have occurred, etc..).

Much more common is the fallacy that you mention, and on which the TVERSKY/KAHNEMAN paper focused.

My fallacy is perhaps more often overlooked by intelligent people who have a strong grasp of statistics, and certainly one whose offenders are much better credentialed (than the general public who is vulnerable to common small number fallacies which you address well here: ).


I personally don't use a lot of statistics, in general, although I am a strong believer in 'profiling'. My point is actually not about that perspective at all.
My point is about being able to see reality beyond the limits of statistics. And this is certainly not a knock on statistic based, or number based, or data based players... That is not my point at all. In fact I would love to put some of the basic things that I look at in unique ways, into a program, which at the very least would save time finding plays and be accurate in finding/not-missing more plays.

There are statistics that are pretty much pure. Maybe these are measuring the flow of capital(pools). There are statistics that are "abstractions"(perhaps a performance figure). Sometimes I can outperform statistics with an abstract basis regardless of sample size.

Lets get back to small numbers and whether we can call them insignificant...
For example, a SUPERtrainer is on an ice cold streak this week...
His FTS ran a dud, his closer couldn't close, his first of the claim didn't move-Up,... his heavy fav. ran 3rd or 4th.
Statistically it means nothing.

Let me try to list some possibilities(i could not care less if I miss some, just go along for the point here)

He is about the same(or even going well), these are randomness
He actually is ice cold.
^ These are not meant to be a 50/50 probability or anything like that, merely an unweighted list.


Now lets look at application:


Wager as usual
or Pass races in which he is in
or Bet against that SuperTrainer with an alternative contender, or spread
If you Pass these races, you cover your ass. You had a "NEGATIVE" and because the sample was small you were not sure if it was real or a false flag. Unlike baseball, you can't get a called strike in horseplaying (The closest equivalent is needing to pay bills, a time constraint/responsibility most casual players don't have.). As long as a large part of your profits or churn don't rely on a certain factor, when that factor has a negative small number sample, you are OK to pass those situations.


Positive or Negative small number samples, may be investigated further. Occasionally you can gather enough info to wager against(or ON -in the case of small number "Positive" samples). These could be discovering statistically significant larger number data that the small number data lead you to. Or these could be areas where your specialized elite insight was outperforming abstract small-number data.






Finally the one that 'provokes' me is the example of a successful player whose success is shown to be statistically irrelevant.



Maybe in some cases, whether or not a person is successful is actually ambiguous, or even a person whose talent, level of insight, and wisdom would actually have a slightly higher probability of being less(or more) successful than they are, yet randomness has favored(or hindered) them in their career/year/week/day...


Yet some people are genuinely elite in their fields. Even (relatively) small numbers can not put any doubt onto their ability.

Hi RF,

You seem to express an affinity for betting based on very recent small streaks - sounds like a belief in the 'hot hand', no? Since Tversky and Kahnemann have appeared in this thread, maybe worth mentioning that they were they people who torpedoed this belief, the opposite of the 'gambler's fallacy', in a study of the shooting of the Boston Celtics nearly thirty years ago.

Cheers,

lansdale

[Dave Schwartz]

You seem to express an affinity for betting based on very recent small streaks - sounds like a belief in the 'hot hand', no? Since Tversky and Kahnemann have appeared in this thread, maybe worth mentioning that they were they people who torpedoed this belief, the opposite of the 'gambler's fallacy', in a study of the shooting of the Boston Celtics nearly thirty years ago.

Yes, we all heard that in the dugout as well.

Of course, the guy on a streak (good or bad) will be hard pressed to agree with those stats.

I think where ALL those studies fall apart is telling the difference between an aberration and a real streak. Not that I can tell the difference either, mind you.

[CincyHorseplayer]

Yes, we all heard that in the dugout as well.

Of course, the guy on a streak (good or bad) will be hard pressed to agree with those stats.

I think where ALL those studies fall apart is telling the difference between an aberration and a real streak. Not that I can tell the difference either, mind you.

I agree and I'll go one step further,any good player is going to streak several times over the course of a year.There seems to be a persistent belief that anybody can be lucky but not good regularly.Talking to countless players on here including yourself I think that is completely false.

[raybo]

Just a few responses-

Hi Raybo,

I meant this as I positive comment about an important aspect of Benter's model that is usually overlooked which has something in common with yours, as you've described it, but you seem a little defensive. I wasn't implying that you could or should want to bet on that scale.

Glad you pointed that out. Global model versus Local models (the one thing Benter and I might have in common) - the hierarchy in my mind would be: global database > Track - 1,2,3, etc. > Race Surface - 1,2,3 etc. > Race Distance - Spr or Rte, or Distance grouping 1,2,3 etc. > Race type Mdn - NonMdn, etc. > Sex M F Open > etc., etc., etc., but, number of subgroups become limited due to resulting sample sizes. If you are track specific, you don't need huge sample sizes and I believe small, recent samples are actually more efficient and predictive in the short term, but must be constantly updated regarding recency, keeping the sample size the same or similar.

Very large samples are great for finding general trends and indicators, but not so great at predicting the short term. Smaller, more time valid samples appear to me to be much more predictive of the near term results.

Benter, for example, was looking for brute force churn leverage, bet a bunch to make a small percentage on that total investment/re-investment. I am looking for near term, consistent cash flow, to support more profit potential vertical exotics play, both of those at a level that allows me to have a relaxed and comfortable life, outside of racing.

The point of my 'multi-millionaire' comment, which is not addressed to you alone, but to *anyone* who has been playing with an edge, or whose model displays positive EV over a sustained period. For anyone who has such an edge, their bank could (or should) be increasing exponentially. Thaskalos made the same point in another thread. Or to paraphrase the great Dick Schmidt, 'If you're playing with a 30% ROI, you don't need a money management system, just a place to stack your money'.

I regularly remove cash from my bankroll before I approach the point at which percentage of bank bet sizing negatively affects odds, and I specialize in small to medium tracks, thus small to medium pool sizes. I also play a small number of tracks whose average payouts are on the high end, which results in limited churn. So, "exponential" bankroll growth is not, nor will it ever be, a goal of mine. I tried that once and though successful overall, the experience was not enjoyable, nor was it healthy, for me or my family.

Now, I understand from your posts here that you are quite risk averse, which, as I said, can seem mysterious, considering the obvious time and energy you've invested in handicapping, but that implies no disrespect. Actually, the opposite. And, as I said, 'different strokes'.

You are absolutely correct. I am not a risk taker, nor do I consider myself a gambler, even though many here think we are all gamblers. Gambling, to me, means risking without a "more than reasonable" expectation of positive returns. I expect and demand profit, and would not participate if the opposite ever became an expectation, so I do not consider myself a gambler or a risk taker.

BTW, no need to respond to the posts of the serial narcissist who frequently pops up on this site to undermine anyone who claims to be making money gambling. As Shakespeare suggested, Iago specializes in bad advice.

I try really hard to ignore obnoxiousness, but I am human and subject to human frailties. One of my points of philosophy, and one that I at times struggle with myself, and that causes me the most angst when relating to others, is my deep belief that we should all act responsibly, respectfully, honestly, and humbly. The fact that others don't believe the same way is my problem, not theirs. The fact that, at times, I find myself acting the way they do, is also my fault, not theirs. None of us is perfect, but we should strive in that direction nevertheless.

Best,

lansdale

[traynor]

Hi Raybo,

I meant this as I positive comment about an important aspect of Benter's model that is usually overlooked which has something in common with yours, as you've described it, but you seem a little defensive. I wasn't implying that you could or should want to bet on that scale.

The point of my 'multi-millionaire' comment, which is not addressed to you alone, but to *anyone* who has been playing with an edge, or whose model displays positive EV over a sustained period. For anyone who has such an edge, their bank could (or should) be increasing exponentially. Thaskalos made the same point in another thread. Or to paraphrase the great Dick Schmidt, 'If you're playing with a 30% ROI, you don't need a money management system, just a place to stack your money'.

Now, I understand from your posts here that you are quite risk averse, which, as I said, can seem mysterious, considering the obvious time and energy you've invested in handicapping, but that implies no disrespect. Actually, the opposite. And, as I said, 'different strokes'.

BTW, no need to respond to the posts of the serial narcissist who frequently pops up on this site to undermine anyone who claims to be making money gambling. As Shakespeare suggested, Iago specializes in bad advice.

Best,

lansdale

One would think that someone as experienced in marketing as Dick Schmidt would be more meticulous in stating such things. It creates the impression that he (or you, if you are misquoting) do not understand the difference between ROI and POI. One would also think that anyone aspiring to expertise in race analysis should be able to catch such obvious errors.

[Robert Fischer]

Hi RF,

You seem to express an affinity for betting based on very recent small streaks - sounds like a belief in the 'hot hand', no? Since Tversky and Kahnemann have appeared in this thread, maybe worth mentioning that they were they people who torpedoed this belief, the opposite of the 'gambler's fallacy', in a study of the shooting of the Boston Celtics nearly thirty years ago.

Cheers,

lansdale

No, I am not a proponent of "hot hand" betting. My insight often extends beyond statistics, to the point that at times I'll notice potential pitfalls related to popular statistical beliefs.

[traynor]

In regard to ROI being "equivalent" to an "advantage" or "edge" of the same degree applied to each and every race, or that a modest ROI derived from what developers call "dirty data" over a clump of races is one short step away from exponential growth, soon to be followed by fortune and glory, I suggest those not conversant in marketing terminology may not recognize the term I mentioned. It is a common in marketing and sales circles:

http://www.urbandictionary.com/define.php?term=treat%20em%20like%20a%20mushroom

The argument goes, "Don't waste time trying to convince them with logic. Treat 'em like mushrooms."

It is actually an interesting spin on the Petty and Cacioppo Elaboration Likelihood Model, usually referred to as the Yale Model of Persuasion.

[traynor]

I want to make clear that my comments are not to insult anyone, but rather to express the opinion that people tend to do better at the track if they are not chasing rainbows. The more accurate information they have, the better they do.

My comments about small samples and deceptive ROIs has nothing to do with anyone on this forum. They are based on a number of years of deep involvement with the Sartin Methodology, and later with All-Ways. Lots of people did the same. Some of them won. Most of them did not. I was one of the latter group. It took me a long time to find out why. I hope to provide information that will shorten the journey for others.

Again, in regard to the topic of this thread, one should understand that an ROI does not equate to an advantage of that amount in every race, nor does it indicate an edge in every race--especially if is based on "dirty data" that has not been meticulously cleaned to eliminate not just a few outliers, but to create a realistic, usable model that is predictive of potential return (or lack of return).

[raybo]

Again, in regard to the topic of this thread, one should understand that an ROI does not equate to an advantage of that amount in every race, nor does it indicate an edge in every race--especially if is based on "dirty data" that has not been meticulously cleaned to eliminate not just a few outliers, but to create a realistic, usable model that is predictive of potential return (or lack of return).

When did you miss the boat! I thought we all agreed with that a long, long time ago, except for your "outliers" opinion. Each to his/her own.

[Robert Fischer]

I was cocky enough as a 24 year old to parlay a year of 229% ROI into an entire winter of 01 of life sustenance into 02.A couple winters of that.Did the same in 2004 successfully.I have bigger dreams but for a kid from nothing I'm allright.

To be continued.

just occurred to me that you meant your $2 ROI.

Makes more sense now.

[traynor]

And on the topic of ROI, I am pleased that my periodic obligatory "put up or shut up" selections posted (on the harness thread--these are, after all, free)
turned out well. Enough for bus fare to Poughkeepsie and lunch tomorrow. At Bellagio's.

Does that count as "doubling the bankroll"?

[Dave Schwartz]

Traynor,

You will need a few thousand more wagers for it to be "meaningful."

(You taught us that.)

:lol:

[traynor]

Traynor,

You will need a few thousand more wagers for it to be "meaningful."

(You taught us that.)

:lol:

I agree wholeheartedly. Small samples are as foolish when they are mine as when they are anyone else's. That was my point. "ROI" figures are essentially meaningless nonsense.

[raybo]

I agree wholeheartedly. Small samples are as foolish when they are mine as when they are anyone else's. That was my point. "ROI" figures are essentially meaningless nonsense.

In your opinion? Or is that a universal fact? :bang:

[lansdale]

In your opinion? Or is that a universal fact? :bang:

Traynor's right. I know you really know this.

[sjk]

Most relevant post in this thread. Very sorry you don't post here more often.

Best,

lansdale

Hi Lansdale,

I am still reading the board every day if not posting. I don't see that many threads that revolve around making a computer program to do the full handicapping process which is the only area where I have experience.

For years I would post a thought that I believed would generate some discussion and often no one replied to my post. I don't think there are many here coming at it from that direction so I have posted little.

To get back to the subject of the thread when I think of ROI I am thinking of actual results over a lengthy stretch of time, a year, 5 years, 10 years. I am pretty sure that there is statistical significance in the results of betting 50,000 races. I don't know why Traynor keeps coming back to "small samples" and associated caveats; no one else is discussing small samples that I can see.

Does a 15 years ROI have relevance to today's pools? Yes and no I think. You can't question the sample size but I do think pools are more efficient (and often smaller) than in years past and the prices are knocked down to an unacceptable level on a lot of what I might have bet in the past.

For me there are fewer playable races (so it is questionable whether sitting by the computer on a weekday is a good use of time) and I am more often on lower probability, higher payout wagers than in the past. For short and medium price horses the visible pools seem efficient much of the time.

Best Wishes.

SK

[lansdale]

Yes, we all heard that in the dugout as well.

Of course, the guy on a streak (good or bad) will be hard pressed to agree with those stats.

I think where ALL those studies fall apart is telling the difference between an aberration and a real streak. Not that I can tell the difference either, mind you.

Hi Dave,

Don't really know what you mean here. Psychologically, it's natural to respond to your present reality. But as you know, all phenomena, including all sports performances, regress to the mean. The great biologist Stephen Jay Gould, who was a huge sports fan, was surprised to find that in all the careers of the greatest baseball players, none of their achievements that he tested fell outside the normal distribution except one - DiMaggio's 56-game hitting streak.

As far as outliers (aberrations) go, you can pick your level of granularity (2 SD, 3 SD?) of length of streak for a given hit rate. I've seen the formula posted on this site a few times. And it's much longer than most players probably think. You've read Taleb - this is his basic point - variance is much greater than most people believe or are even capable of understanding - in fact, it's a bitch.

Cheers,

lansdale

[limit2]

I see a 40% ROI is possible on the Win slot. Perhaps it can best be done with longshots and a moderate progression. Best of luck to those who try to achieve this. Thank you.

[lansdale]

Hi Lansdale,

I am still reading the board every day if not posting. I don't see that many threads that revolve around making a computer program to do the full handicapping process which is the only area where I have experience.

For years I would post a thought that I believed would generate some discussion and often no one replied to my post. I don't think there are many here coming at it from that direction so I have posted little.

To get back to the subject of the thread when I think of ROI I am thinking of actual results over a lengthy stretch of time, a year, 5 years, 10 years. I am pretty sure that there is statistical significance in the results of betting 50,000 races. I don't know why Traynor keeps coming back to "small samples" and associated caveats; no one else is discussing small samples that I can see.

Does a 15 years ROI have relevance to today's pools? Yes and no I think. You can't question the sample size but I do think pools are more efficient (and often smaller) than in years past and the prices are knocked down to an unacceptable level on a lot of what I might have bet in the past.

For me there are fewer playable races (so it is questionable whether sitting by the computer on a weekday is a good use of time) and I am more often on lower probability, higher payout wagers than in the past. For short and medium price horses the visible pools seem efficient much of the time.

Best Wishes.

SK

Hi sjk,

Thanks for posting. I take your point on the non-response to your posts. But I think that there are very few people here who take your approach to the game and have your level of expertise - a math PhD. Magister Ludi is one. There are also probably few who have developed a model that has been profitable for many years. So I think I and others have much to gain from your advice, although I realize the reverse is not often the case.

I think many players are witnessing the increasing efficiency in mutuel pools in recent years and are responding in a similar manner.

Since there's so much discussion of sample size, I think your take on the minimum sample size required for a robust, stable model would be very helpful. Bill Benter and a few others have mentioned the range of 2500-3000 as being sufficient. You refer to 50k as possibly reaching too far into the past, but what would your opinion be on this?

Much thanks for your input, as always.

Best,

lansdale

[sjk]

Formula_2002 used to have a formula on his website (this is many years ago) where you could put in the frequency of wins and the average mutual along with the sample size and it would give a interval of confidence for your results.

For a reasonably frequent hit rate (say 10%) I would have to think the 2500-3000 would be sufficient to draw conclusions. Perhaps Formula_2002 or someone else who knows statistics could provide this formula.

As Traynor points out you do not want to apply the formula the day after you make a big score but rather for some arbitrarily chosen block of races.

[traynor]

Whatever the sample size, consistency is probably the most desirable attribute for betting. With smaller samples (of whatever size) it may be useful to run the results through a bootstrapping algorithm. Excel may have one, and if not, it is readily available. The random sampling with replacement will quickly determine if the results are consistently spread across the sample (indicating the small sample may be predictive of distribution across a larger sample), or whether there are a few high peaks among mostly valleys.

[DeltaLover]

We have had similar discussions many times so far and of course there is no
doubt about the need of extensive historical data to approach the problem from a
pure analytical view. By this I mean that we usually start with the intention of
somehow deriving a probability distribution for each starter of the race which
will make it possible to discover overlays that will be used for the creation of
a betting execution.

This is not the only approach though.

We can also attack the problem from a strictly technical analysis point of view,
where our objective is shifted from a pure analytical view, to a more practical
approach where we try to create scripts showing profitability based in several
dimensions (circuit and time might be two of them) while we allow them to be
mutable as new data are coming in to the system. Of course, I realize that such
an approach might not have much of academic merit, but experience can prove that
it is still valid.

To make my statement more apparent, I refer you to a comparison to a stock
market trading model using fundamentals (this can be Benter's approach or
anything similar) against a technical analysis based model (which would be the
mutable scripts).

[traynor]

We have had similar discussions many times so far and of course there is no
doubt about the need of extensive historical data to approach the problem from a
pure analytical view. By this I mean that we usually start with the intention of
somehow deriving a probability distribution for each starter of the race which
will make it possible to discover overlays that will be used for the creation of
a betting execution.

This is not the only approach though.

We can also attack the problem from a strictly technical analysis point of view,
where our objective is shifted from a pure analytical view, to a more practical
approach where we try to create scripts showing profitability based in several
dimensions (circuit and time might be two of them) while we allow them to be
mutable as new data are coming in to the system. Of course, I realize that such
an approach might not have much of academic merit, but experience can prove that
it is still valid.

To make my statement more apparent, I refer you to a comparison to a stock
market trading model using fundamentals (this can be Benter's approach or
anything similar) against a technical analysis based model (which would be the
mutable scripts).

Well said. I do this for income, not for glory or academic purposes. That makes my primary focus on finding stuff that works. As you point out above, that may differ from an approach that has "academic merit"--but may be more profitable. The first rule should be "Get the money." Academic merit and the accolades of others can wait until some later date.

[traynor]

Exactly so. Add to that the fact that regardless of how many races an individual handicaps, on a grand scale it is no more than an infinitesimal part of a much larger population. Extrapolating from a severely limited group in a sub-sample that one has uncovered Deep and Meaningful Truths about the nature of reality in horse racing is a manifestation of ego and hubris rather than acquired wisdom. I refer anyone who believes otherwise to the comments by Schlesinger on the "statistically meaningless" sample represented by Anderson's blackjack play.

I defer--again--to Mr. Schlesinger's expertise and opinion.

[Robert Fischer]

I see a 40% ROI is possible on the Win slot. Perhaps it can best be done with longshots and a moderate progression. Best of luck to those who try to achieve this. Thank you.

I'm not sure if this is a reference to someone showing a 40% win ROI (if so, and relevant, please point it out), but as a general answer; 40% win ROI is what I would say falls under the umbrella of "spot plays".
Whether that means longshots(that you legitimately actually happen to 'see' as contenders), or more popular public choices(they still have to be overlays however to get an average of 40%), ..., 40% win ROI is going to represent a successful run on your 'select' plays.
This isn't a beauty contest, (or even an ROI contest), so any superstar level ROIs are going to be, -in my opinion, representative of inefficient play.

check the following quote, feel free to debate or expand :ThmbUp:
[[red], added now by me]

A lot of people online seem to have ROIs of 1.5,[1.4 (40%)], 1.2, etc... The first thing that comes to mind after filtering out those that I can actually picture beating the game and keeping somewhat accurate records, is how inefficient such a high ROI is. It may sound funny, but a pro or semi-pro who has been playing for a good amount of time, and is hitting those kind of ROI numbers, is going to want to start bringing those ROIs down. It's just the way the pools, and the rebates work. With the anomalies affecting the ROI, those can go both ways, but it's probably far more common online to hear of inflated ROIs than deflated.

[raybo]

Traynor's right. I know you really know this.

Traynor seems to not know how to state things in ways that make those statements true, universally, and yet he states them as being true universally. I was being facetious in my response. First he states that small samples are foolish (I take that as meaning it is universally true, which it is not, but Traynor fails to qualify it that way), second he states that ROI is meaningless nonsense (again I take that as meaning it is universally true, which it is not, and again he fails to qualify it that way).

If you choose to believe that small samples can never be of any value whatsoever, then that's up to you. And, if you choose to believe that ROI has absolutely no value, again, that's up to you also. You would be deluding yourself however, IMO. I use small samples successfully, and I also use ROI, in combination with other factors, successfully. So, if I use them both, successfully, and I'm pretty certain there are others who do also, then neither statement he made could possibly be universally true, because he used no qualifiers in those statements, and that, my friend, is just not good enough, when you're putting it out there as being fact.

Modeling is used by many players, and when you do that you end up with several models of small sample sizes. According to Traynor those people are "foolish". Many people measure their success with ROI, at the end of meets, at the end of the year, 2 years, 5 years, etc., and according to Traynor all those ROIs are "meaningless and nonsense".

Now, do you agree, or not? I certainly do not agree. And, I have the proof in my bankroll, and my ADW account to prove it.

[baconswitchfarm]

I think everyone can have their opinion on this. But when you dictate your opinion as facts it ruffles people. I have seen consistent six figure earners year after year, be told on here they may not be successful. They have just fooled themselves into thinking that because their stats don't run to infinity. I won't live to infinity so I don't need my stats to project that far.

[Robert Fischer]

But as you know, all phenomena, including all sports performances, regress to the mean.
This is a great way to put it.

What someone looking at small sample streaks..(hothands/getting in the zone,etc..) should really be doing is looking at whether or not the "mean" itself has shifted.

The exchange from posts 109,138 (http://www.paceadvantage.com/forum/showpost.php?p=1522180&postcount=138),143 (http://www.paceadvantage.com/forum/showpost.php?p=1522395&postcount=143) was driving at this topic.

back to your quote: ... phenomena...regress to the mean.

In order for a horseplayer to accurately see that the "mean itself" has shifted, there are some requirements. I have tried to produce a decent list here that hopefully is accurate enough for the purpose of this thread:



data based proof. = BIG MONEY The data shows that there is enough probability to suggest that mean has shifted (or, the reversal: in spite of the public wagering as though a streak or "hot hand" is taking place, the data shows such a streak to be improbable, and thus provides an advantage).
Significant Changes (handlers,health,equipment,etc).
Inside Information.
The ability to see changes in the mean through a combination of data and observational techniques.
other


I'm a "#4".(Truth be told, I'm not exclusively a #4, but it plays some part in my edge, and more significantly serves a point). Size does matter :blush:... I am by no means dismissing sample size. I can say that my edge over data increases when it is both an area that I am strong in, and an area in which the data has a degree of abstraction. I can also add that passively abstaining from possible negatives, works within this system, more safely than actively backing possible positives.

I also completely understand the dismissal and disbelief from a "#1", -at least until the sample size reaches a certain level. That doubt or disbelief is not foreign to me at this point. In fact, I received that same dismissal/disbelief yesterday, from a psychiatrist (albeit behind the world's worst 'poker face'), while explaining to him that I am reasonably intelligent, during our introductory meeting.[if this "yarn" doesn't boost my internet cred', I don't know what will! ;)].


couple quick semantic issues:


To think that I take the same data, and ever dismiss sample size, is a misunderstanding of my perspective.
my quote from post #105 (http://www.paceadvantage.com/forum/showpost.php?p=1520808&postcount=105), boils down to the importance of not dismissing a statistically insignificant sample as "insignificant". -"insignificant" leaves the field of statistics.
Usually things that seem like anomalies, actually fall within (, and closer to the heart of,) the normal range of the "mean".

[MJC922]

Hi Lansdale,

I am still reading the board every day if not posting. I don't see that many threads that revolve around making a computer program to do the full handicapping process which is the only area where I have experience.

For years I would post a thought that I believed would generate some discussion and often no one replied to my post. I don't think there are many here coming at it from that direction so I have posted little.

To get back to the subject of the thread when I think of ROI I am thinking of actual results over a lengthy stretch of time, a year, 5 years, 10 years. I am pretty sure that there is statistical significance in the results of betting 50,000 races. I don't know why Traynor keeps coming back to "small samples" and associated caveats; no one else is discussing small samples that I can see.

Does a 15 years ROI have relevance to today's pools? Yes and no I think. You can't question the sample size but I do think pools are more efficient (and often smaller) than in years past and the prices are knocked down to an unacceptable level on a lot of what I might have bet in the past.

For me there are fewer playable races (so it is questionable whether sitting by the computer on a weekday is a good use of time) and I am more often on lower probability, higher payout wagers than in the past. For short and medium price horses the visible pools seem efficient much of the time.

Best Wishes.

SK

It would be nice to take the human element out of it. I'm not convinced that it's viable though to substantially profit from computer selections. Don't get me wrong it's worth investigating as I have no interest in putting myself back into the emotional meat grinder of fulltime horseplay anymore. I find the subtleties of trips, pace and other visual observations still bring clear value at times... this is to be expected as these are largely invisible in the surface data which is now all but mere fodder for whaling at this point. The corrections the bots are making are to the point where they're laughably predictable and if we still have any price left when they get done with it often it's because the horse is dead on the board.

[raybo]

The corrections the bots are making are to the point where they're laughably predictable and if we still have any price left when they get done with it often it's because the horse is dead on the board.

This should be a clue for you, regarding automated systems.

[MJC922]

This should be a clue for you, regarding automated systems.

Undeniably many people know what's largely predictive in the game and are crunching it -- the same data you're probably crunching too. I don't think automation is a problem in this equation but I could be wrong.

[raybo]

Undeniably many people know what's largely predictive in the game and are crunching it -- the same data you're probably crunching too. I don't think automation is a problem in this equation but I could be wrong.

Could you explain that a bit, I'm not sure I understand your exact meaning, or perhaps I misinterpreted the statement I quoted.

[lansdale]

Traynor seems to not know how to state things in ways that make those statements true, universally, and yet he states them as being true universally. I was being facetious in my response. First he states that small samples are foolish (I take that as meaning it is universally true, which it is not, but Traynor fails to qualify it that way), second he states that ROI is meaningless nonsense (again I take that as meaning it is universally true, which it is not, and again he fails to qualify it that way).

If you choose to believe that small samples can never be of any value whatsoever, then that's up to you. And, if you choose to believe that ROI has absolutely no value, again, that's up to you also. You would be deluding yourself however, IMO. I use small samples successfully, and I also use ROI, in combination with other factors, successfully. So, if I use them both, successfully, and I'm pretty certain there are others who do also, then neither statement he made could possibly be universally true, because he used no qualifiers in those statements, and that, my friend, is just not good enough, when you're putting it out there as being fact.

Modeling is used by many players, and when you do that you end up with several models of small sample sizes. According to Traynor those people are "foolish". Many people measure their success with ROI, at the end of meets, at the end of the year, 2 years, 5 years, etc., and according to Traynor all those ROIs are "meaningless and nonsense".

Now, do you agree, or not? I certainly do not agree. And, I have the proof in my bankroll, and my ADW account to prove it.

Hi Raybo,

Do you remember a book I mentioned a few months ago, Maboussin's "The Success Equation: How to Tell the Difference Between Skill and Luck in Sports and Business", by the chief investment officer of Legg-Mason?" When we discussed it, you seemed to agree with me that Maboussin's fundamental premise, that it is only by means of a sufficiently large sample size can one determine the difference between luck and skill in any performance. Another way of referring to this is the 'Law of Large Numbers'. It is universal. There's no way of getting around that. Traynor didn't create it, so let's leave him out of this. Now you seem to say you accept it, but only sometimes. You must realize that this doesn't make sense.

First, let's put aside the ROI debate. It's a trivial semantic quibble about a term on which most gamblers (or investors) agree. For the sake of this post, I'm using the blackjack term EV (expected value) to refer to either the bettor's advantage (or lack of same) with respect to either an individual bet or any bet sequence of indeterminate length.

Although you keep accurate records and have obviously done a great deal of testing, your play is not automated - your model is not a black box. You're ultimately a subjective handicapper, like Thaskalos. Thus, your EV can only be determined in retrospect, from a large sample of races. And that would include whatever decisions you make based on smaller samples. What the EV of those small samples may be is insignificant - what matters is your global EV, over, let's say a 2500 race sample. Obviously, since you've been profitable for a number of years, this is positive.

You may say the small samples are statistically significant, I say they're variance. You may say, 'Then why am I so profitable'. Like you, I can only conjecture. My guess (which I've already mentioned elsewhere) is that using track-specific models is part of it, but would also add that, in my experience, and in the demonstrated results of people like Benter, both the characteristics of individual tracks and the characteristics of their betting publics remain extremely stable over time. And I would conjecture that this is also true for your track models, which although you might analyze them on a seasonal basis, likely show little change over a larger sample of recent years. I would also conjecture that the ongoing changes you make likely have much less effect on your global EV than you think.

So, I would say, in sum, that small samples are useless.

Cheers,

lansdale

[lansdale]

This is a great way to put it.

What someone looking at small sample streaks..(hothands/getting in the zone,etc..) should really be doing is looking at whether or not the "mean" itself has shifted.

The exchange from posts 109,138 (http://www.paceadvantage.com/forum/showpost.php?p=1522180&postcount=138),143 (http://www.paceadvantage.com/forum/showpost.php?p=1522395&postcount=143) was driving at this topic.

back to your quote:

In order for a horseplayer to accurately see that the "mean itself" has shifted, there are some requirements. I have tried to produce a decent list here that hopefully is accurate enough for the purpose of this thread:


data based proof. = BIG MONEY The data shows that there is enough probability to suggest that mean has shifted (or, the reversal: in spite of the public wagering as though a streak or "hot hand" is taking place, the data shows such a streak to be improbable, and thus provides an advantage).
Significant Changes (handlers,health,equipment,etc).
Inside Information.
The ability to see changes in the mean through a combination of data and observational techniques.
other
I'm a "#4".(Truth be told, I'm not exclusively a #4, but it plays some part in my edge, and more significantly serves a point). Size does matter :blush:... I am by no means dismissing sample size. I can say that my edge over data increases when it is both an area that I am strong in, and an area in which the data has a degree of abstraction. I can also add that passively abstaining from possible negatives, works within this system, more safely than actively backing possible positives.

I also completely understand the dismissal and disbelief from a "#1", -at least until the sample size reaches a certain level. That doubt or disbelief is not foreign to me at this point. In fact, I received that same dismissal/disbelief yesterday, from a psychiatrist (albeit behind the world's worst 'poker face'), while explaining to him that I am reasonably intelligent, during our introductory meeting.[if this "yarn" doesn't boost my internet cred', I don't know what will! ;)].


couple quick semantic issues:


To think that I take the same data, and ever dismiss sample size, is a misunderstanding of my perspective.
my quote from post #105 (http://www.paceadvantage.com/forum/showpost.php?p=1520808&postcount=105), boils down to the importance of not dismissing a statistically insignificant sample as "insignificant". -"insignificant" leaves the field of statistics.
Usually things that seem like anomalies, actually fall within (, and closer to the heart of,) the normal range of the "mean".

Hi Robert,

There may be a disconnect here. I don't know about others, but when I am referring to 'mean' in this context, I'm talking about the mean of a betting model's or bettor's over all (global) advantage. How can one race 'shift' this?

Cheers,

lansdale

[raybo]

Hi Raybo,

Do you remember a book I mentioned a few months ago, Maboussin's "The Success Equation: How to Tell the Difference Between Skill and Luck in Sports and Business", by the chief investment officer of Legg-Mason?" When we discussed it, you seemed to agree with me that Maboussin's fundamental premise, that it is only by means of a sufficiently large sample size can one determine the difference between luck and skill in any performance. Another way of referring to this is the 'Law of Large Numbers'. It is universal. There's no way of getting around that. Traynor didn't create it, so let's leave him out of this. Now you seem to say you accept it, but only sometimes. You must realize that this doesn't make sense.

First, let's put aside the ROI debate. It's a trivial semantic quibble about a term on which most gamblers (or investors) agree. For the sake of this post, I'm using the blackjack term EV (expected value) to refer to either the bettor's advantage (or lack of same) with respect to either an individual bet or any bet sequence of indeterminate length.

Although you keep accurate records and have obviously done a great deal of testing, your play is not automated - your model is not a black box. You're ultimately a subjective handicapper, like Thaskalos. Thus, your EV can only be determined in retrospect, from a large sample of races. And that would include whatever decisions you make based on smaller samples. What the EV of those small samples may be is insignificant - what matters is your global EV, over, let's say a 2500 race sample. Obviously, since you've been profitable for a number of years, this is positive.

You may say the small samples are statistically significant, I say they're variance. You may say, 'Then why am I so profitable'. Like you, I can only conjecture. My guess (which I've already mentioned elsewhere) is that using track-specific models is part of it, but would also add that, in my experience, and in the demonstrated results of people like Benter, both the characteristics of individual tracks and the characteristics of their betting publics remain extremely stable over time. And I would conjecture that this is also true for your track models, which although you might analyze them on a seasonal basis, likely show little change over a larger sample of recent years. I would also conjecture that the ongoing changes you make likely have much less effect on your global EV than you think.

So, I would say, in sum, that small samples are useless.

Cheers,

lansdale

No, I don't remember ever saying that large samples are needed, and my method is a back box. All I do is make sure the minimum odds are met and place the wagers. My samples are small because they are track specific, and further divided by 20 PPG categories, for eliminations only. The remainder of the contenders are determined and ranked by velocities.

[lansdale]

No, I don't remember ever saying that large samples are needed, and my method is a back box. All I do is make sure the minimum odds are met and place the wagers. My samples are small because they are track specific, and further divided by 20 PPG categories, for eliminations only. The remainder of the contenders are determined and ranked by velocities.

Hi Raybo,

Okay. You've decided to suspend the laws of probability. :-). At least now we know where you stand.

Cheers,

lansdale

[raybo]

Hi Raybo,

Okay. You've decided to suspend the laws of probability. :-). At least now we know where you stand.

Cheers,

lansdale

I've had the same stance for a long time Lansdale, it is well known on this board, and I'm still here, and still making money. My personal philosophy is that large samples can be used for spotting general possibilities and trends, nothing more, unless one drills down into those larger samples to get at what is important at each track and surface/distance. That's been my approach all along, and no talk of large samples and the laws of probability, or of large numbers, is going to change that. Do what makes you money and I'll do what makes me money, we're all different people, and different players, working from different philosophies, having different goals and priorities.

I could put thousands of races in my database, but to what end? Dilution would result, lowering my profitability, probably to such a degree that I would not be profitable anymore. I'm unwilling to let that happen. I currently have 2 years of data for lots of tracks, but I only use the most recent data in my individual track databases, because I want to know how the tracks have been playing recently, not what they played like 2 years ago, or 5 years ago, etc.. Does that sound illogical to you? If so, then we have nothing left to discuss on this subject.

[DeltaLover]

I've had the same stance for a long time Lansdale, it is well known on this board, and I'm still here, and still making money. My personal philosophy is that large samples can be used for spotting general possibilities and trends, nothing more, unless one drills down into those larger samples to get at what is important at each track and surface/distance. That's been my approach all along, and no talk of large samples and the laws of probability, or of large numbers, is going to change that. Do what makes you money and I'll do what makes me money, we're all different people, and different players, working from different philosophies, having different goals and priorities.

There's more than one way to skin a cat.

I think this applies here. Although what you say about small samples is not a valid assumption from a pure academic approach this does not necessary mean that in the real world a method like yours is impossible to be profitable. In this game all it matters is showing profit, the why this is happening is of secondary or even non importance at all. As far as your system is able to change frequently and does not remain static, I believe it has a substantial probability to become profitable.

[raybo]

There's more than one way to skin a cat.

I think this applies here. Although what you say about small samples is not a valid assumption from a pure academic approach this does not necessary mean that in the real world a method like yours is impossible to be profitable. In this game all it matters is showing profit, the why this is happening is of secondary or even non importance at all. As far as your system is able to change frequently and does not remain static, I believe it has a substantial probability to become profitable.

My system changes every time a race is recorded into the database for that track, however slightly that change may be, it is updated by each race's winner. But, again, the database is only used for win contender eliminations, not for selections other than that the remaining horses remain contenders.

[traynor]

Small samples don't "prove" anything--other than that something happened in a very small number of cases--that can be usefully applied to predict future events. That is not a "prejudice" against small samples. It is a simple, provable fact of reality. The smaller the sample, the more unlikely the results are to be predictive when applied to a larger sample.

In the real world, the experience(s) of one person, or of a small group of persons, do not extrapolate to Greater Truth for the entire world. In the real world, the experience(s) of one person, or of a small group of persons are nothing more than the experience(s) of one person, or of a small group of persons. Those experiences, however interesting, are in no way predictive. Perhaps in lala land it is acceptable to believe they are predictive, but the truth is that they are not--they only describe a very limited subset of events that create the illusion of a "pattern."

The reason people use small samples, and advocate small samples, is simple. Larger samples would establish that the premises derived from the small samples are nonsense.

[thaskalos]

The winning percentage of the race favorites has remained remarkably stable through the years, and their ROI had remained very stable as well, up 'til the recent "whale" era. It is my contention that the steadiness in the performance of these favorites has been the driving force in fostering many players' belief that their own performance as bettors in this game could also attain the same level of consistency.

I haven't found that to be the case in my own play.

My own betting results fluctuate wildly month-to-month, and also fluctuate greatly year-to-year...to the extent where I hesitate to state with certainty what my ROI really is. It seems to always be in a state of flux...and yet to be determined.

[raybo]

Small samples don't "prove" anything--other than that something happened in a very small number of cases--that can be usefully applied to predict future events. That is not a "prejudice" against small samples. It is a simple, provable fact of reality. The smaller the sample, the more unlikely the results are to be predictive when applied to a larger sample.

In the real world, the experience(s) of one person, or of a small group of persons, do not extrapolate to Greater Truth for the entire world. In the real world, the experience(s) of one person, or of a small group of persons are nothing more than the experience(s) of one person, or of a small group of persons. Those experiences, however interesting, are in no way predictive. Perhaps in lala land it is acceptable to believe they are predictive, but the truth is that they are not--they only describe a very limited subset of events that create the illusion of a "pattern."

The reason people use small samples, and advocate small samples, is simple. Larger samples would establish that the premises derived from the small samples are nonsense.

Ok, Traynor, we get it! I'll just throw all my winnings in the trash and start all over. Is that what you want? Will you be satisfied that your great knowledge is recognized then? Geez!!