Making your own handicapping system.

[Capper Al]

"The formulation of the problem is often more essential than its solution, which may be merely a matter of mathematical or experimental skill." Albert Einstein

Oldman (my handicapping partner of sorts) and I often discuss the madness to our methods.Oldman has a database and handicaps with that and does well. I haven't built my database yet. I have spent 5 or more years writing a C++ program that feeds large spreadsheets, one for maidens and another for non-maidens. Oldman and I approach solving the handicapping riddle from opposite ends of the do-it yourself system cappers.

There have been many posts recommending the sensibility of record keeping and finding your own niche in the game. And yes I totally agree. But the DIY handicapper doesn't have the resourses of the commercial software vendor that sponsor PA, so the individual on his own has to choose their method of attack on how to tool up to play the game. Our time becomes our limited resource.

Having and using a database does give one the power of Percentages and Probabilities answering the most valued racing question of what happens in this situation. But I grew up in the paper and pencil period of handicapping before personal computers. I loved generating figures with a calculator for speed or pace or the dot system and many more other systems. In doing this I came to believe that the riddle to handicapping was best solved in the alchemy of the formulas being used. Now I persist in this direction as I'm about to rewrite my system instead of build my database. Why? Because I only have time to pursue one path. I believe from writing my original application that now I understand the game more and can build a better application. After my rewrite, I will build my database.

[HUSKER55]

first off, you can buy data to build a database so I think creating an application you are content with to be the most important.

If you always feel "wanting" then something is wrong and you "know it". you might not know what, but you know something is amiss.

that is why I keep records of where my application hits and misses and why.

One day....I will hit the big one! :D

good luck with your endeavors

[traynor]

"The formulation of the problem is often more essential than its solution, which may be merely a matter of mathematical or experimental skill." Albert Einstein

Oldman (my handicapping partner of sorts) and I often discuss the madness to our methods.Oldman has a database and handicaps with that and does well. I haven't built my database yet. I have spent 5 or more years writing a C++ program that feeds large spreadsheets, one for maidens and another for non-maidens. Oldman and I approach solving the handicapping riddle from opposite ends of the do-it yourself system cappers.

There have been many posts recommending the sensibility of record keeping and finding your own niche in the game. And yes I totally agree. But the DIY handicapper doesn't have the resourses of the commercial software vendor that sponsor PA, so the individual on his own has to choose their method of attack on how to tool up to play the game. Our time becomes our limited resource.

Having and using a database does give one the power of Percentages and Probabilities answering the most valued racing question of what happens in this situation. But I grew up in the paper and pencil period of handicapping before personal computers. I loved generating figures with a calculator for speed or pace or the dot system and many more other systems. In doing this I came to believe that the riddle to handicapping was best solved in the alchemy of the formulas being used. Now I persist in this direction as I'm about to rewrite my system instead of build my database. Why? Because I only have time to pursue one path. I believe from writing my original application that now I understand the game more and can build a better application. After my rewrite, I will build my database.

Databases can be as misleading as they can be informative. That is, raw data is just that--data rather than information. Your approach may be best. Figure out what information you need first, then build the database to provide you with that information. That will avoid the all-too-common problem of continual "fishing expeditions" that turn up imaginary trends in the raw data that only exist when viewed in small clumps and small samples.

I use a "database" but it is comprised of handicapped races--not raw data. Meaning I have automated the process of "handicapping" individual races and use the output of that application as input for the data mining apps. It works a LOT better than chasing rainbows in raw data. For one thing, because the handicapping app automatically cleans the data (including the elimination of outliers), the information that results from the data mining process is substantially more reliable, and rarely misleading. That is a lot more than can be said of apps massaging raw, uncleaned data.

[DRIVEWAY]

Do you have criteria to determine whether a race is playable? I'm not talking about identifying contenders and searching for odds value, I'm talking about prior to any wagering answering the question - "Is this race a playable beatable type of race".

If you need handicappers to evaluate and comment on your specifications or testers to work your program and provide feedback, let me know. It's fun collaborating on these projects.

Good Luck in your endeavor.

[Robert Goren]

I am not a data base handicapper, but something I have always been curious about is how you handle track bias in your data. Do you ignore them? Throw races that have them? Have separate data bases for the most common?

[DeltaLover]

"The formulation of the problem is often more essential than its solution, which may be merely a matter of mathematical or experimental skill." Albert Einstein

Oldman (my handicapping partner of sorts) and I often discuss the madness to our methods.Oldman has a database and handicaps with that and does well. I haven't built my database yet. I have spent 5 or more years writing a C++ program that feeds large spreadsheets, one for maidens and another for non-maidens. Oldman and I approach solving the handicapping riddle from opposite ends of the do-it yourself system cappers.

There have been many posts recommending the sensibility of record keeping and finding your own niche in the game. And yes I totally agree. But the DIY handicapper doesn't have the resourses of the commercial software vendor that sponsor PA, so the individual on his own has to choose their method of attack on how to tool up to play the game. Our time becomes our limited resource.

Having and using a database does give one the power of Percentages and Probabilities answering the most valued racing question of what happens in this situation. But I grew up in the paper and pencil period of handicapping before personal computers. I loved generating figures with a calculator for speed or pace or the dot system and many more other systems. In doing this I came to believe that the riddle to handicapping was best solved in the alchemy of the formulas being used. Now I persist in this direction as I'm about to rewrite my system instead of build my database. Why? Because I only have time to pursue one path. I believe from writing my original application that now I understand the game more and can build a better application. After my rewrite, I will build my database.

The fact that you are using C++ to generate spreadsheets while your friend
prefers a data base does not necessary mean that you are approaching handicapping
from opposite ends. The difference relies on the methodology and not on the
mechanisms used.

I am not sure that a database gives one the power of percentages and
probabilities as you say here. This seems like a very primitive approach to the
process of computerized handicapping. Before you will be in the position to
maintain any valid expectation for some positive results, who need to address
challenges like improving data quality, defining a solid theoretical foundation
of meta handicapping, getting familiar with applicable domains of arterial
intelligence and then follow up with a long series of trial and error
experiments. Let me give you an example: You say that you love generating
figures, have you ever think that instead of creating them based on your
experience it might be possible to develop an algorithm to allow the computer to
derive a whole universe of indexes and use them for handicapping? More than this
you need also to understand how to make a judgement call about the quality and
effectiveness of each index and figure and how to use it more effectively,
generating a probability distribution based on it, a ranking sequence of
anything else.

[HUSKER55]

delta lover, I read your post and maybe I am in a rut and don't realize it. [not to worry...this isn't the first rodeo] :D

can you give an example using speed or pace. Thinking here is it should be easy to understand what you mean from the example.

But alas, maybe I am wrong.

thanks

[DeltaLover]

delta lover, I read your post and maybe I am in a rut and don't realize it. [not to worry...this isn't the first rodeo] :D

can you give an example using speed or pace. Thinking here is it should be easy to understand what you mean from the example.

But alas, maybe I am wrong.

thanks


Hope the following is helpful:

Let's assume that we have a specific methodology to calculate some sort of a
speed figure (similar to Beyer or Brawn). For this step we need the algorithm to
derive this figure. Think as an example Beyer's method. Note that we have
infinite ways to model the past performance data to a figure, so the computer
can be used here, to automatically create its logic.


Each individual figure is not of much of a value until we somehow compose them
to a rating (similar to bris prime power) that can describe each horse with
a unique number. Again as before there are infinite ways to create the rating,
among other factors, we can consider recency, earnings, connections etc to
estimate the weight of each individual figure.

After we construct this rating for each horse the next step is to convert it to
some scale that will be easy to back test. One obvious way is for example to
create a probability distribution based on them. This process can again be
assisted by the computer by applying a genetic program for example.

The last step is to take these probabilities and backtest them through
historical data to see how well the fit. For example we might try to see how
well this probabilities predict the winner. The way to do so, is another model,
that can be as simple as comparing it against the final odds for example. This
comparison will later allow us to compare two set of figures and decide which
one is better than the other, or in other words it has better fitness.

Note, that this approach is not limited to the winning horse and speed figures
only. We can create for example another set of figures trying to predict who
will be on the lead in the first call of the race or even what is the horse that
will finish last.

[Capper Al]

Databases can be as misleading as they can be informative. That is, raw data is just that--data rather than information. Your approach may be best. Figure out what information you need first, then build the database to provide you with that information. That will avoid the all-too-common problem of continual "fishing expeditions" that turn up imaginary trends in the raw data that only exist when viewed in small clumps and small samples.

I use a "database" but it is comprised of handicapped races--not raw data. Meaning I have automated the process of "handicapping" individual races and use the output of that application as input for the data mining apps. It works a LOT better than chasing rainbows in raw data. For one thing, because the handicapping app automatically cleans the data (including the elimination of outliers), the information that results from the data mining process is substantially more reliable, and rarely misleading. That is a lot more than can be said of apps massaging raw, uncleaned data.

Thanks fellow developer. I appreciate this coming especially from you. This is my belief exactly. Take this as a simple item to deal with: I have an algorithm that handles what to do if there isn't a good BRIS speed figure available for a PP line. Just the data cleaning situations are necessary, besides all the complication of handicapping theory.

[Capper Al]

The fact that you are using C++ to generate spreadsheets while your friend
prefers a data base does not necessary mean that you are approaching handicapping
from opposite ends. The difference relies on the methodology and not on the
mechanisms used.

I am not sure that a database gives one the power of percentages and
probabilities as you say here. This seems like a very primitive approach to the
process of computerized handicapping. Before you will be in the position to
maintain any valid expectation for some positive results, who need to address
challenges like improving data quality, defining a solid theoretical foundation
of meta handicapping, getting familiar with applicable domains of arterial
intelligence and then follow up with a long series of trial and error
experiments. Let me give you an example: You say that you love generating
figures, have you ever think that instead of creating them based on your
experience it might be possible to develop an algorithm to allow the computer to
derive a whole universe of indexes and use them for handicapping? More than this
you need also to understand how to make a judgement call about the quality and
effectiveness of each index and figure and how to use it more effectively,
generating a probability distribution based on it, a ranking sequence of
anything else.

You right we could and do have some similar approaches, but I don't look at raw data as much as he does for calculations.

[Capper Al]

first off, you can buy data to build a database so I think creating an application you are content with to be the most important.

If you always feel "wanting" then something is wrong and you "know it". you might not know what, but you know something is amiss.

that is why I keep records of where my application hits and misses and why.

One day....I will hit the big one! :D

good luck with your endeavors

My confidence is in my approach. I have designed and developed several major applications for the Department of Defense. What I mostly know is that I have been profitable each year for the past 5 years, so something is working. Now to be honest had I spent my time flipping hamburgers at McDonalds instead of developing a system and playing horses, I would have earned 20 times plus more money.

[Capper Al]

I am not a data base handicapper, but something I have always been curious about is how you handle track bias in your data. Do you ignore them? Throw races that have them? Have separate data bases for the most common?

Track bias for today's current race is almost impossible to deal with in computation, although I do have a method for it but rarely use it. The hope is over the long run it will wash out for databases. I do calculate bias from race line PP's in my routine calculations. That's different.

[HUSKER55]

NOW you sound like me. There are days that I swear the biggest problem is making the proper wager at the proper time.


Delta lover, thanks for your example.

[Exotic1]

Hope the following is helpful:

Let's assume that we have a specific methodology to calculate some sort of a
speed figure (similar to Beyer or Brawn). For this step we need the algorithm to
derive this figure. Think as an example Beyer's method. Note that we have
infinite ways to model the past performance data to a figure, so the computer
can be used here, to automatically create its logic.


Each individual figure is not of much of a value until we somehow compose them
to a rating (similar to bris prime power) that can describe each horse with
a unique number. Again as before there are infinite ways to create the rating,
among other factors, we can consider recency, earnings, connections etc to
estimate the weight of each individual figure.

After we construct this rating for each horse the next step is to convert it to
some scale that will be easy to back test. One obvious way is for example to
create a probability distribution based on them. This process can again be
assisted by the computer by applying a genetic program for example.

The last step is to take these probabilities and backtest them through
historical data to see how well the fit. For example we might try to see how
well this probabilities predict the winner. The way to do so, is another model,
that can be as simple as comparing it against the final odds for example. This
comparison will later allow us to compare two set of figures and decide which
one is better than the other, or in other words it has better fitness.

Note, that this approach is not limited to the winning horse and speed figures
only. We can create for example another set of figures trying to predict who
will be on the lead in the first call of the race or even what is the horse that
will finish last.

Helpful to me. Thanks.

[keenang]

I remember years ago at a Dr.Sartin seminar he said no 2 races are never.never,ever run exactly the same. So I said to myself why keep a data base and to this day I think he was right.

Gene K. :)

[DeltaLover]

I remember years ago at a Dr.Sartin seminar he said no 2 races are never.never,ever run exactly the same. So I said to myself why keep a data base and to this day I think he was right.

Gene K. :)

This is a nice way of thinking... :rolleyes:

[Fingal]

I remember years ago at a Dr.Sartin seminar he said no 2 races are never.never,ever run exactly the same. So I said to myself why keep a data base and to this day I think he was right.

Gene K. :)

I would modify this. I keep an early energy profile for various distances & then see if my contenders fall within an acceptable deviation from that number. Close does count in more than horse shoes & handgrenades.

True no 2 races are ever the same, but they can have similarities that carry over to future ones. I'm just not going to become Capt. Ahab chasing that great white whale of the elusive exact best number.

[shouldacoulda]

The difference relies on the methodology and not on the
mechanisms used.

Truer words were never spoken. I'm no computer programmer but I would think you would need a winning "system" (and I hate to use that word in relation to horse racing) first and then translate it into a program the computer can interpret. There is so much that remains a judgment call in my eyes I don't know if I would trust a computer to make all my selection decisions. There are so many variables that can influence the outcome of a race. I think it would be hard to compensate for all or even most in a program.

One of the posters on this board once said "each race is an entity unto itself, never to be repeated again". You could have the same horses on the same track with the same jockeys on another day and get a different outcome the majority of the time. I like using angles with race shape and class the most over all indicators. Some people like pace and/or speed figures. It's whatever works for you is the right way.

There are some good programs on the market already. I would think it would be easier to find one that works for you and exploit it to its fullest extent. I wish you luck with your endeavor. It is a noble and colossal undertaking.

[BIG HIT]

Just decision in approach is never ending.How many pp line's to go back what you read is really true or partial truth or if true just dozen't fit angle which my self favor right or wrong then which one's work for you connection owner trainer and thier stats by the time you sort threw what works your right back two where you started is today the day since most hdcper's pick more loser than winners can be just never ending.
And this board has some top hdcper's on it with know how and data base insight i don't think there is a approach that has not been mention.In my opinion hope can find the holy grail by putting more effort in class trainer good luck in your effort's

[Capper Al]

There are many systems out there for sale. Races are more similar than dissimilar when grouped together by race type than not. If racing was hopelessly dissimilar then no ones system would work.

[raybo]

I find that databases are best used for eliminations, not for selections. Eliminate the non-contenders using the database, then compare the contenders against each other, and the synthesis of the field as a whole. Value is the final answer for betting selections.

[Show Me the Wire]

There are many systems out there for sale. Races are more similar than dissimilar when grouped together by race type than not. If racing was hopelessly dissimilar then no ones system would work.

I disagree. The profile of the type of horse is consistent, not the race itself.

[Capper Al]

I disagree. The profile of the type of horse is consistent, not the race itself.

How are you meaning this? How does one know the value of a horse in a race without knowing the field?

[Show Me the Wire]

How are you meaning this? How does one know the value of a horse in a race without knowing the field?

I don't understand your meaning of value without knowing the field.

What I mean by the profile of the type of horse is just that. It depends on the class levels of the circuit snd the eligibility conditions of the race. As an extreme example the profile of a claiming 12.5k nw2l winner is not a winner of a recent MSW race.

[Show Me the Wire]

Capper Al:

After reading my initial response, I believe it is inadequate. A better answer would be is the winner's profile the type of animal that is moving up in class, dropping down in class or remaining at the same class level for the race conditions. That is the basic profile. The basic profile is not dependent on running style, pace, energy distribution, distance, etc.

[Capper Al]

Capper Al:

After reading my initial response, I believe it is inadequate. A better answer would be is the winner's profile the type of animal that is moving up in class, dropping down in class or remaining at the same class level for the race conditions. That is the basic profile. The basic profile is not dependent on running style, pace, energy distribution, distance, etc.

I'm with you on this.

[jasperson]

Al, I know you are using a program now, so how is the new one goinging to be different?

[Capper Al]

Al, I know you are using a program now, so how is the new one goinging to be different?

The orginal system was based on my reading and handicapping expertise at the time I started to code. As I learned more I could modify a lot, but some basic structures would not lend themself to easily be updated. A lot of these modification are do with pace and class. These are the building of many a system as they are for me.

[Dave Schwartz]

PMFJI, but I have an opinion about "building a handicapping system."

Many of us have databases of some significance. If we take a strictly statistical view of horse handicapping, we are in direct competition with the best handicapping engines on the planet (i.e. whales). This includes the Deltas and Tifecta Mike's of our forum.

IMHO, we cannot compete head-to-head with them, using approaches that are highly similar to theirs. IOW, if what we want to do is duplicate their approaches (i.e. Benter, Wagner, etc.) it is highly likely that we will come up short. They have more expertise and more man-power.

So, what is the answer?

Obviously, the answer is different for everyone. It will be a combination of one's desires, skill set and limitations.

I think one must see themselves as being in the same boat as a Mom and Pop business trying to compete in the age of mega chain stores: you have to find a niche that the big guys either cannot or will not exploit.

How do you do this?

I believe there are a variety of ways. The famous baseball quote from Wee Willie Keeler, "Hit it where they ain't" is a good start.

And where ain't they? Well, they're not on price horses, for one place. If they were, these horses would not pay big prices.

Note: While value is certainly the key for whales, their modeling techniques are primarily global in nature. Therefore, their systems push winners to the top. Their very model suggests that the horses must be reasonably low-to-mid-priced, otherwise significant money could not be wagered on them.

("Significant money" is based upon whale-scale.)


Think of this as guerilla warfare. Hit them and run. Be a small target, because you are outnumbered. (Think of dollars as soldiers; we have much smaller armies.)

Ask yourself, "What can we, the small wagering company do that the mega-conglomerate cannot?" One answer is, "Know thy track." This means, allow LOCAL MODELS to drive your handicapping. Go for the smaller, more precise model rather than the giant GLOBAL model. Allow yourself to uncover things that cannot easily be seen with the naked eye from the 50,000 foot level.

Another is "know the trainers." Another is, "Know this race shape." Still another could be "know this tote board." The list of potential factors and angles goes on and on.

Just my opinion.

Regards,
Dave Schwartz

[DeltaLover]

I completely agree with what Dave calls local models. “Localization” can be viewed as an additional layer of specification where grouping by several attributes can have significant impact to the behavior of a model. This concept fits well within a global model as far as it implements a mechanism to discover this type of a cluster.

To be more specific, the track where the race is run, can be seen as another race-level attribute (similar to classification or average recency for example) and be treated as such applying the same methodogy and approach. We can go a step further, allowing subgrouping of the attributes, in the case of track, we can for example conclude that short meeting, top level tracks (like Saratoga, Del Mar and Keenland) are behaving similar and the same might apply for another cluster of low quality tracks (like Phily, Pen and Finger Lakes for example) but we can also verify that a specific track (for example Monmouth) stands alone and need its own, specialized model.

[raybo]

PMFJI, but I have an opinion about "building a handicapping system."

Many of us have databases of some significance. If we take a strictly statistical view of horse handicapping, we are in direct competition with the best handicapping engines on the planet (i.e. whales). This includes the Deltas and Tifecta Mike's of our forum.

IMHO, we cannot compete head-to-head with them, using approaches that are highly similar to theirs. IOW, if what we want to do is duplicate their approaches (i.e. Benter, Wagner, etc.) it is highly likely that we will come up short. They have more expertise and more man-power.

So, what is the answer?

Obviously, the answer is different for everyone. It will be a combination of one's desires, skill set and limitations.

I think one must see themselves as being in the same boat as a Mom and Pop business trying to compete in the age of mega chain stores: you have to find a niche that the big guys either cannot or will not exploit.

How do you do this?

I believe there are a variety of ways. The famous baseball quote from Wee Willie Keeler, "Hit it where they ain't" is a good start.

And where ain't they? Well, they're not on price horses, for one place. If they were, these horses would not pay big prices.

Note: While value is certainly the key for whales, their modeling techniques are primarily global in nature. Therefore, their systems push winners to the top. Their very model suggests that the horses must be reasonably low-to-mid-priced, otherwise significant money could not be wagered on them.

("Significant money" is based upon whale-scale.)


Think of this as guerilla warfare. Hit them and run. Be a small target, because you are outnumbered. (Think of dollars as soldiers; we have much smaller armies.)

Ask yourself, "What can we, the small wagering company do that the mega-conglomerate cannot?" One answer is, "Know thy track." This means, allow LOCAL MODELS to drive your handicapping. Go for the smaller, more precise model rather than the giant GLOBAL model. Allow yourself to uncover things that cannot easily be seen with the naked eye from the 50,000 foot level.

Another is "know the trainers." Another is, "Know this race shape." Still another could be "know this tote board." The list of potential factors and angles goes on and on.

Just my opinion.

Regards,
Dave Schwartz

I sure am glad you're part of this forum! I could not have worded it nearly as well as you just did. Great analogies, and IMO, exactly on point! Well done!!! :jump:

[raybo]

I completely agree with what Dave calls local models. “Localization” can be viewed as an additional layer of specification where grouping by several attributes can have significant impact to the behavior of a model. This concept fits well within a global model as far as it implements a mechanism to discover this type of a cluster.

To be more specific, the track where the race is run, can be seen as another race-level attribute (similar to classification or average recency for example) and be treated as such applying the same methodogy and approach. We can go a step further, allowing subgrouping of the attributes, in the case of track, we can for example conclude that short meeting, top level tracks (like Saratoga, Del Mar and Keenland) are behaving similar and the same might apply for another cluster of low quality tracks (like Phily, Pen and Finger Lakes for example) but we can also verify that a specific track (for example Monmouth) stands alone and need its own, specialized model.

I also agree with you. Where we differ, it appears, is that, in my opinion, a specific race belongs to a specific track model not a global model. When you group similar races from totally different tracks, those race group specificities get diluted to a large extent, and you lose much of what makes races highly similar, or dissemilar. So, you end up making comparisons that don't really exist. In short, individual race groupings, or categories, are track specific/track attributes, not the other way around, IMHO.

Can one utilize a "global" system to get at those "local" models/groupings? Yes, of course, but the next step down in the global model's sub-grouping hierarchy should be "track". That is, of course, if you want to distance yourself from the larger, more damaging effects of the public as a whole, and of course the whales' influence and competition. If you're constantly ending up with lower priced horses, then you're "up against it" so to speak. You will fight that battle against overwhelming armies and your chance of succeeding is quite low, if not impossible. Rebates are really your only savior.

Just my 2 cents.

[thaskalos]

I don't know what other horseplayers do or don't do...so I can't offer any direct advice on how to become a better player, or create a better system. But I have played this game for a very long time...and I know what has hindered me the most in my pursuit to become a winning player. I am in no way unique from other players...so I must assume that the problems that hindered me must also trouble other players out there as well.

The strange thing about being a losing player is that you really won't know what you were doing wrong -- or what your weaknesses were as a player -- until you have transformed yourself into a winning player. As long as you remain a losing player, it is easy to delude yourself about what you are doing wrong...and the changes that you often think that you need to make in your gameplan are often not what your game really needs.

It's the classic catch-22 situation; you know that you are losing money and that you should make some changes in your play...but, as a losing player, you are ill-equipped to properly analyze your situation, and make the right changes in order to take your game to the level it needs to be.

I spend more time at OTBs than the average player spends...and I see this scenario play itself out on a regular basis. Many losing players tell me that they would easily become winners if they could just be able to infuse some "discipline" into their game...seemingly oblivious to the fact that the vast majority of the "disciplined" horseplayers lose too, albeit to a lesser extent. You have to be a WINNER in order to know what it really takes to win...because then you'll be speaking from actual experience. When a losing player talks about what it takes to win, he is only hypothesizing...based on what he has read or heard from others. And there is no way of knowing if what he heard or read is actually true.

Do we really need more knowledge or a better "system"...or are we just incapable of playing the game to the best of our ability, no matter what "system" we follow?

Is it the GAME that's really beating us...or is the game just giving us the opportunity to beat ourselves, and we happily oblige?

We will only know if we find out for ourselves. The most important knowledge is "self-knowledge".

This is an unforgiving game, and it magnifies whatever character flaws we might have, to the extent where we become our own worst enemies at the track. And no one knows this better than I.

Success at the track entails so much more than just having the "right" system...

[mountainman]

PMFJI, but I have an opinion about "building a handicapping system."

Many of us have databases of some significance. If we take a strictly statistical view of horse handicapping, we are in direct competition with the best handicapping engines on the planet (i.e. whales). This includes the Deltas and Tifecta Mike's of our forum.

IMHO, we cannot compete head-to-head with them, using approaches that are highly similar to theirs. IOW, if what we want to do is duplicate their approaches (i.e. Benter, Wagner, etc.) it is highly likely that we will come up short. They have more expertise and more man-power.

So, what is the answer?

Obviously, the answer is different for everyone. It will be a combination of one's desires, skill set and limitations.

I think one must see themselves as being in the same boat as a Mom and Pop business trying to compete in the age of mega chain stores: you have to find a niche that the big guys either cannot or will not exploit.

How do you do this?

I believe there are a variety of ways. The famous baseball quote from Wee Willie Keeler, "Hit it where they ain't" is a good start.

And where ain't they? Well, they're not on price horses, for one place. If they were, these horses would not pay big prices.

Note: While value is certainly the key for whales, their modeling techniques are primarily global in nature. Therefore, their systems push winners to the top. Their very model suggests that the horses must be reasonably low-to-mid-priced, otherwise significant money could not be wagered on them.

("Significant money" is based upon whale-scale.)


Think of this as guerilla warfare. Hit them and run. Be a small target, because you are outnumbered. (Think of dollars as soldiers; we have much smaller armies.)

Ask yourself, "What can we, the small wagering company do that the mega-conglomerate cannot?" One answer is, "Know thy track." This means, allow LOCAL MODELS to drive your handicapping. Go for the smaller, more precise model rather than the giant GLOBAL model. Allow yourself to uncover things that cannot easily be seen with the naked eye from the 50,000 foot level.

Another is "know the trainers." Another is, "Know this race shape." Still another could be "know this tote board." The list of potential factors and angles goes on and on.

Just my opinion.

Regards,
Dave Schwartz

Inspired post, Dave. You're at your best in light computer- armor. The entire suit is too constricting for the man inside to flex and communicate his considerable knowledge.

I would add that whale programs seem better at weighing merit than at flagging negative form traits of a subtle nature. A seasoned human with the mental scars of losing wagers can vet situations for potential pitfalls on an almost subtextual level that no machine can tap into.

I also believe that the computer between a man's ears is best at downloading results. That's because, while a machine swallows raw input, however specific in nature, the mind of a really good handicapper UNDERSTANDS what it just witnessed and refines it accordingly before filing it away.

Prognostications aside, ask a computer to RECAP a race sometime, that is, to seize on WHY the perhaps unpredictable happened-to explain what's apparent after-the-fact..that, for instance, some longshot was sent on a suicide mission to engage the favorite early; that the 8-1 winner shared some seemingly obscure form-trait with another recent winner from the same outfit; that the winner unexpectedly reverted to a running-style it had suceeded with in the past..etc etc. Wouldn't a computer program's blindspots, lack of "feel," that is, hinder it on BOTH sides of the outcome? Before AND after?

Much respect, as always, and would appreciate your feedback.

[raybo]

I don't know what other horseplayers do or don't do...so I can't offer any direct advice on how to become a better player, or create a better system. But I have played this game for a very long time...and I know what has hindered me the most in my pursuit to become a winning player. I am in no way unique from other players...so I must assume that the problems that hindered me must also trouble other players out there as well.

The strange thing about being a losing player is that you really won't know what you were doing wrong -- or what your weaknesses were as a player -- until you have transformed yourself into a winning player. As long as you remain a losing player, it is easy to delude yourself about what you are doing wrong...and the changes that you often think that you need to make in your gameplan are often not what your game really needs.

It's the classic catch-22 situation; you know that you are losing money and that you should make some changes in your play...but, as a losing player, you are ill-equipped to properly analyze your situation, and make the right changes in order to take your game to the level it needs to be.

I spend more time at OTBs than the average player spends...and I see this scenario play itself out on a regular basis. Many losing players tell me that they would easily become winners if they could just be able to infuse some "discipline" into their game...seemingly oblivious to the fact that the vast majority of the "disciplined" horseplayers lose too, albeit to a lesser extent. You have to be a WINNER in order to know what it really takes to win...because then you'll be speaking from actual experience. When a losing player talks about what it takes to win, he is only hypothesizing...based on what he has read or heard from others. And there is no way of knowing if what he heard or read is actually true.

Do we really need more knowledge or a better "system"...or are we just incapable of playing the game to the best of our ability, no matter what "system" we follow?

Is it the GAME that's really beating us...or is the game just giving us the opportunity to beat ourselves, and we happily oblige?

We will only know if we find out for ourselves. The most important knowledge is "self-knowledge".

This is an unforgiving game, and it magnifies whatever character flaws we might have, to the extent where we become our own worst enemies at the track. And no one knows this better than I.

Success at the track entails so much more than just having the "right" system...

Not much I can add to this - great post! Of course, the first thing we must do is conquer ourselves. Fortunately, I possessed, whatever you want to call it, before I ever started studying racing. I knew going in that I could do it, but didn't know how exactly, so I began studying the game, as an "external/non-participant", looking down from above (that's the best explanation I can think of). Starting from that vantage point enabled me to see the "big picture" with no preconceived notions, other than that I would be successful, eventually. I had no bad habits to unlearn, no bad advice to forget, etc.. I had no boundaries preset for me, so I could widen my focus to include everything that I was capable of seeing.

I think we agree that there is one "overriding" factor in racing, and gambling in general, one either has the personal traits needed to succeed, or one doesn't. The "system" we use in our success evolves over time, it may start out with the DRF and a pencil, and stay there, or one can evolve into a computer "assisted" system, and some even evolve into a computer "driven" system, with very little user input. But, regardless of the system one uses in his/her overall methodology, one must first be "personally capable" of succeeding. Unfortunately, many are not capable of succeeding, no matter what methodology they employ. So, they are in a constant struggle to find the right system, when in fact, the system is not the problem, and it's certainly not the answer. The answer is either already within us, or it isn't. If one "doesn't have it" they can certainly improve themselves as players, but it is doubtful they will ever become successful players, in a strictly monetary sense anyway (note I didn't say "impossible", I said "doubtful").

[dkithore]

[QUOTE=Dave Schwartz]PMFJI, but I have an opinion about "building a handicapping system."

Many of us have databases of some significance. If we take a strictly statistical view of horse handicapping, we are in direct competition with the best handicapping engines on the planet (i.e. whales). This includes the Deltas and Tifecta Mike's of our forum.

IMHO, we cannot compete head-to-head with them, using approaches that are highly similar to theirs. IOW, if what we want to do is duplicate their approaches (i.e. Benter, Wagner, etc.) it is highly likely that we will come up short. They have more expertise and more man-power.

So, what is the answer?

Obviously, the answer is different for everyone. It will be a combination of one's desires, skill set and limitations.

I think one must see themselves as being in the same boat as a Mom and Pop business trying to compete in the age of mega chain stores: you have to find a niche that the big guys either cannot or will not exploit.

How do you do this?

I believe there are a variety of ways. The famous baseball quote from Wee Willie Keeler, "Hit it where they ain't" is a good start.

And where ain't they? Well, they're not on price horses, for one place. If they were, these horses would not pay big prices.

Note: While value is certainly the key for whales, their modeling techniques are primarily global in nature. Therefore, their systems push winners to the top. Their very model suggests that the horses must be reasonably low-to-mid-priced, otherwise significant money could not be wagered on them.

("Significant money" is based upon whale-scale.)


Think of this as guerilla warfare. Hit them and run. Be a small target, because you are outnumbered. (Think of dollars as soldiers; we have much smaller armies.)

Ask yourself, "What can we, the small wagering company do that the mega-conglomerate cannot?" One answer is, "Know thy track." This means, allow LOCAL MODELS to drive your handicapping. Go for the smaller, more precise model rather than the giant GLOBAL model. Allow yourself to uncover things that cannot easily be seen with the naked eye from the 50,000 foot level.

Another is "know the trainers." Another is, "Know this race shape." Still another could be "know this tote board." The list of potential factors and angles goes on and on.

Just my opinion.

Regards,
Dave Schwartz[/QUOTE
very worthwhile points to ponder for pragmetic players.

[traynor]

Of all the many. many approaches I tried over the years--often with short-term success and long-term disaster--the single factor that made the most difference in my bottom line was to stop chaisng rainbows.

Specifically, i used models, but those models were distorted by simple averaging of mutuel prices in calculating ROI. I can trace the start of consistent, long-term winning to a very simple process.

Put all the mutuels from whatever approach you are using into an array.
Sort the array from smallest to biggest (mutuel price).
Divide the length of the array (the number of elements) by 4. Call that n.
(Convert n to an integer if it is a double.)
Pick the nth element in the array. (One quarter of the length from the start.)
Pick the -nth element in the array. (One quarter of the length from the end.)
Add the two element values together, and divide by two. Then multiply by 1.5. Call that MIQR.
Cycle through the array of mutuel prices again, with conditional:

If value greater than MIQR, then value = MIQR.

If your handicapping shows a positive ROI when so cleaned, you really have something.

That simple step made me realize that in MANY of the cases in which I thought I had an advantage, I did not. That simple step made me realize that in MANY of the cases in which I thought I had an edge, I did not. The upside is that instead on congratulating myself on the astuteness of my race analysis skills, I kept looking.

The value of that step is such that I would never consider wagering based on the output of any software application that uses simple averaging, or some slipshod, half-baked method of "correcting for outliers." The downside is that I have to spend a lot of time working on my own apps.

The upside is that when this dreary city is buried in snow, I will be in Miami, thoroughly enjoying my winter, all expenses paid by the kind folk who bet on all those other horses.

[Dave Schwartz]

The value of that step is such that I would never consider wagering based on the output of any software application that uses simple averaging, or some slipshod, half-baked method of "correcting for outliers." The downside is that I have to spend a lot of time working on my own apps.

Whose software does that?

[raybo]

Whose software does that?

Heck, everyone knows that the only races that count are in the 1st and last quarter. And you gotta use those 2 to get rid of all the payouts you had that were too high, right? Just because you bet a horse that had good value doesn't mean anything, if they hit it's just luck, never to be repeated.

[whodoyoulike]

Of all the many. many approaches I tried over the years--often with short-term success and long-term disaster--the single factor that made the most difference in my bottom line was to stop chaisng rainbows.

Specifically, i used models, but those models were distorted by simple averaging of mutuel prices in calculating ROI. I can trace the start of consistent, long-term winning to a very simple process.

Put all the mutuels from whatever approach you are using into an array.
Sort the array from smallest to biggest (mutuel price).
Divide the length of the array (the number of elements) by 4. Call that n.
(Convert n to an integer if it is a double.)
Pick the nth element in the array. (One quarter of the length from the start.)
Pick the -nth element in the array. (One quarter of the length from the end.)
Add the two element values together, and divide by two. Then multiply by 1.5. Call that MIQR.
Cycle through the array of mutuel prices again, with conditional:

If value greater than MIQR, then value = MIQR.

If your handicapping shows a positive ROI when so cleaned, you really have something.

That simple step made me realize that in MANY of the cases in which I thought I had an advantage, I did not. That simple step made me realize that in MANY of the cases in which I thought I had an edge, I did not. The upside is that instead on congratulating myself on the astuteness of my race analysis skills, I kept looking.

The value of that step is such that I would never consider wagering based on the output of any software application that uses simple averaging, or some slipshod, half-baked method of "correcting for outliers." The downside is that I have to spend a lot of time working on my own apps.

The upside is that when this dreary city is buried in snow, I will be in Miami, thoroughly enjoying my winter, all expenses paid by the kind folk who bet on all those other horses.


When you divide by 2 then multiply by 1.5, isn't that the same as multiplying your sum by .75? Why the two step process? Why 1.5? What is MIQR? Thanks,

[dnlgfnk]

"The formulation of the problem is often more essential than its solution, which may be merely a matter of mathematical or experimental skill." Albert Einstein

Oldman (my handicapping partner of sorts) and I often discuss the madness to our methods.Oldman has a database and handicaps with that and does well. I haven't built my database yet. I have spent 5 or more years writing a C++ program that feeds large spreadsheets, one for maidens and another for non-maidens. Oldman and I approach solving the handicapping riddle from opposite ends of the do-it yourself system cappers.

There have been many posts recommending the sensibility of record keeping and finding your own niche in the game. And yes I totally agree. But the DIY handicapper doesn't have the resourses of the commercial software vendor that sponsor PA, so the individual on his own has to choose their method of attack on how to tool up to play the game. Our time becomes our limited resource.

Having and using a database does give one the power of Percentages and Probabilities answering the most valued racing question of what happens in this situation. But I grew up in the paper and pencil period of handicapping before personal computers. I loved generating figures with a calculator for speed or pace or the dot system and many more other systems. In doing this I came to believe that the riddle to handicapping was best solved in the alchemy of the formulas being used. Now I persist in this direction as I'm about to rewrite my system instead of build my database. Why? Because I only have time to pursue one path. I believe from writing my original application that now I understand the game more and can build a better application. After my rewrite, I will build my database.

Two things...(1) I obtain accurate percentages as well as anyone else by simply applying public odds to the race at hand. (2) The best handicapper race in, race out (public) uses obvious data to achieve that consistency.

The paradigm shift that needs to occur is the transformation from..."This 3-2 favorite wins the greatest percentage of the time", to..."This 3-2 favorite loses 67% of the time".

[Capper Al]

I don't know what other horseplayers do or don't do...so I can't offer any direct advice on how to become a better player, or create a better system. But I have played this game for a very long time...and I know what has hindered me the most in my pursuit to become a winning player. I am in no way unique from other players...so I must assume that the problems that hindered me must also trouble other players out there as well.

The strange thing about being a losing player is that you really won't know what you were doing wrong -- or what your weaknesses were as a player -- until you have transformed yourself into a winning player. As long as you remain a losing player, it is easy to delude yourself about what you are doing wrong...and the changes that you often think that you need to make in your gameplan are often not what your game really needs.

It's the classic catch-22 situation; you know that you are losing money and that you should make some changes in your play...but, as a losing player, you are ill-equipped to properly analyze your situation, and make the right changes in order to take your game to the level it needs to be.

I spend more time at OTBs than the average player spends...and I see this scenario play itself out on a regular basis. Many losing players tell me that they would easily become winners if they could just be able to infuse some "discipline" into their game...seemingly oblivious to the fact that the vast majority of the "disciplined" horseplayers lose too, albeit to a lesser extent. You have to be a WINNER in order to know what it really takes to win...because then you'll be speaking from actual experience. When a losing player talks about what it takes to win, he is only hypothesizing...based on what he has read or heard from others. And there is no way of knowing if what he heard or read is actually true.

Do we really need more knowledge or a better "system"...or are we just incapable of playing the game to the best of our ability, no matter what "system" we follow?

Is it the GAME that's really beating us...or is the game just giving us the opportunity to beat ourselves, and we happily oblige?

We will only know if we find out for ourselves. The most important knowledge is "self-knowledge".

This is an unforgiving game, and it magnifies whatever character flaws we might have, to the extent where we become our own worst enemies at the track. And no one knows this better than I.

Success at the track entails so much more than just having the "right" system...

I know this to be true when I'm in a losing streak. I frantically search for what I'm doing wrong or what recent change in my code caused the losing situation. And then somehow it comes to an end and I start winning again. Sometimes I'll have a clue as to why and sometimes I won't have a clue as to what happened.

[Capper Al]

The question here is really how we use technology. Our starting point tells much about who we are and if we are to be successful. There are basically to approaches when we start. One is from stats and the other from handicapping. I do believe that those starting from stats are chasing rainbows. A computer is best utilized by processing formulas much more faster and accurate then we as humans can possibly do. Stats should follow our computations to measure how well we are doing, but our computations should not be derived from our stats. This is mainly because horse racing is an open system where we just don't know the finite possibilities like a die having 6 sides. We cannot truly say that in today's race that in this situation this horse will win 3 out of 5 times because we cannot verify the situation as similar.

[DeltaLover]

The question here is really how we use technology. Our starting point tells much about who we are and if we are to be successful. There are basically to approaches when we start. One is from stats and the other from handicapping. I do believe that those starting from stats are chasing rainbows. A computer is best utilized by processing formulas much more faster and accurate then we as humans can possibly do. Stats should follow our computations to measure how well we are doing, but our computations should not be derived from our stats. This is mainly because horse racing is an open system where we just don't know the finite possibilities like a die having 6 sides. We cannot truly say that in today's race that in this situation this horse will win 3 out of 5 times because we cannot verify the situation as similar.

I think that the question is more about meta handicapping theory rather than how to use the technology, which can be considered just an implementation detail. More than this, what I find more interesting when it comes to betting, is not that much the ability of a computer to accelerate to quickly process a formula but its ability to discover it as well.

[jasperson]

[QUOTE=One is from stats and the other from handicapping. I do believe that those starting from stats are chasing rainbows. A computer is best utilized by processing formulas much more faster and accurate then we as humans can possibly do. Stats should follow our computations to measure how well we are doing, but our computations should not be derived from our stats. This is mainly because horse racing is an open system where we just don't know the finite possibilities like a die having 6 sides. We cannot truly say that in today's race that in this situation this horse will win 3 out of 5 times because we cannot verify the situation as similar.[/QUOTE]
Al, you lost me here. How do we handicap without using stats? Anytime we look at pp to handicap a race we are using stats When we look at a horse's pp and he ran well at this level in his last race chances are he will run well in this race. How do we know that? It's from stats. To those who say you can't use averages I say yes you can Insurance companies do it very successfully.

[HUSKER55]

TRAYNOR, be a sport and post an example. keep it simple as I am not the sharpest knife in the drawer.


thanks! :)

[raybo]

The question here is really how we use technology. Our starting point tells much about who we are and if we are to be successful. There are basically to approaches when we start. One is from stats and the other from handicapping. I do believe that those starting from stats are chasing rainbows. A computer is best utilized by processing formulas much more faster and accurate then we as humans can possibly do. Stats should follow our computations to measure how well we are doing, but our computations should not be derived from our stats. This is mainly because horse racing is an open system where we just don't know the finite possibilities like a die having 6 sides. We cannot truly say that in today's race that in this situation this horse will win 3 out of 5 times because we cannot verify the situation as similar.

That's why I stated earlier that a database, developing stats/percentages, is best used for eliminations, not for selections. Let the stats/percentages tell you which horses are "non-contenders", then use your handicapping, whether computer generated or long hand via human handicapping, tell you which of the remaining contenders is the likely winner, then let value tell you when to bet on that likely winner.

[raybo]

Al, you lost me here. How do we handicap without using stats? Anytime we look at pp to handicap a race we are using stats When we look at a horse's pp and he ran well at this level in his last race chances are he will run well in this race. How do we know that? It's from stats. To those who say you can't use averages I say yes you can Insurance companies do it very successfully.

I agree, we use stats, whether formally, or informally, depending on the method we use, technical or non-technical. And I also agree that averages can be used successfully, but those averages must be from valid data, filtered for the specifics of the race at hand. Averages of averages of averages, etc., are probably not the way to go. One must start with legitimate specifics and hone/filter those specifics before averaging. IMO.

[Capper Al]

Al, you lost me here. How do we handicap without using stats? Anytime we look at pp to handicap a race we are using stats When we look at a horse's pp and he ran well at this level in his last race chances are he will run well in this race. How do we know that? It's from stats. To those who say you can't use averages I say yes you can Insurance companies do it very successfully.

We must have a base feeling for what's going on. We need to understand the game somehow before we proceed to handicap. How we get there doesn't matter, but once we decide to use technology we must select an approach to start with, and this usually comes down to querying data or creating fomulas like the old paper and pencil players did. All I'm saying that I believe creating the formulas is the way to start.

[traynor]

When you divide by 2 then multiply by 1.5, isn't that the same as multiplying your sum by .75? Why the two step process? Why 1.5? What is MIQR? Thanks,

The process is from code. The two-step process is just that--to illustrate that it is in two steps. The two values are added together, then averaged. That value is multiplied by 1.5. The value of 1.5 is that it represents a reasonably accurate model of future potential. "ROI" on past races using dirty data is misleading, because it imples that the small sample is representative of a larger population--which it usually is not.

MIQR is an acronym for mean of the interquartile range.

If you run your results through such a cleaning process and it shows a positive ROI, you may have something worthwhile. If the results cannot show a profit using cleaned data, it is fairly safe to assume the results are based on anomalies that will not be replicated when applied to another group of races.

[traynor]

TRAYNOR, be a sport and post an example. keep it simple as I am not the sharpest knife in the drawer.


thanks! :)

Say you have a sample of 100 races. If the mutuel prices are sorted from lowest to highest, the lowest may be 2.20 and the highest 89.00. The purpose of modeling is predicting the future. That purpose can best be fulfilled by using an approximation that is far more likely to be repeated than simple averaging of the 100 mutuel prices would indicate.

Using the list of sorted mutuel prices, pick the 25th from the bottom and the 25th from the top. Those values may be 8.00 and 4.00. Add to get 12.00. Divide by 2 to get 6.00--which is an accurate predictive value of what future average prices will be.

Multiply the 6.00 by 1.5 to get 9.00. Set that as a default value to replace any mutuel price greater than that amount and go through the list of mutuel prices again using:

If (mutuel price) > 9.00 then (mutuel price) = 9.00

Use the new cleaned data to calculate an ROI. If it is positive, you may have something that can usefully be applied to future races. Not guaranteed, but a much more dependable model than using dirty data.

BTW, the MIQR is used routinely in predictive computer modeling and simulations. It is not some personal quirk of mine.

[JJMartin]

I agree, we use stats, whether formally, or informally, depending on the method we use, technical or non-technical. And I also agree that averages can be used successfully, but those averages must be from valid data, filtered for the specifics of the race at hand. Averages of averages of averages, etc., are probably not the way to go. One must start with legitimate specifics and hone/filter those specifics before averaging. IMO.

Ray, when betting your final contenders, do you use post time odds or m/l? Also you said you run each track through 11 different systems and choose the best performing one at the time. Is the choice according to ROI or highest strike rate? How many races back do you test each method?

[whodoyoulike]

... That value is multiplied by 1.5. The value of 1.5 is that it represents a reasonably accurate model of future potential. "ROI" on past races using dirty data is misleading, because it imples that the small sample is representative of a larger population--which it usually is not.

MIQR is an acronym for mean of the interquartile range.

If you run your results through such a cleaning process and it shows a positive ROI, you may have something worthwhile. If the results cannot show a profit using cleaned data, it is fairly safe to assume the results are based on anomalies that will not be replicated when applied to another group of races.

I'm still unclear, why are you are using a factor of 1.5? How do you know that it represents a "reasonably accurate model of future potential"?

Thanks,

[raybo]

Ray, when betting your final contenders, do you use post time odds or m/l? Also you said you run each track through 11 different systems and choose the best performing one at the time. Is the choice according to ROI or highest strike rate? How many races back do you test each method?

Since I'm requiring a minimum odds to justify bets, I am using gate time odds, the last click on the tote before they enter the gate. I also watch the odds up to that time and anticipate the direction the odds will go after the race has started.

We use a combination of ROI and hit rate and number of plays, and we also look at the listing of recorded plays to get an idea of longest losing streaks and frequency of losing streaks. It's a personal preference for each user, some want many plays (action or churn) and are willing to sacrifice a bit in the other factors, some don't mind passing races and will choose a method that offers fewer plays but a higher hit rate and/or ROI, etc.. Much is dependent on the track, some give you several profitable methods to choose from, others only 1 profitable method, and some tracks simply will not produce a profit, usually lower average win payout tracks, like Emd or CrC, or tracks that play really weird, like Mnr and TuP, etc..

I normally test all the past cards I have, back to 2011- 2012 if possible. But, for the longer meets, 3 months or more, we can put the first couple of months in the database and test the next month or so, before playing. There are about half the methods that do not require a database for eliminations, so for those methods one can begin testing with the first day of a meet and after about 3 or 4 weeks decide which of those methods is performing best. We always record all the plays everyday, even if we don't play every day, so we can track how our method is doing, as well as all the other methods, and sometimes we find that things have changed at the track and we need to switch methods.

We have a testing process called "All Record" that records every race on every card that qualifies as a play in all 11 methods, with the click of a single button, so we can always look at each method and see how it is doing, at any point in time. We can also look at every race that was played in each method, at any point in time as they are all listed on each methods record keeping sheet, along with the hit rates, ROI, low payout, high payout, average payout, and median payout, total dollars bet and total dollars paid, for all individual betting interests as well as any combination of those 3 possible betting interests. And, of course we have a summary of those factors listed by minimum odds ranges from .01/1 through 11/1 and higher.

[Capper Al]

Say you have a sample of 100 races. If the mutuel prices are sorted from lowest to highest, the lowest may be 2.20 and the highest 89.00. The purpose of modeling is predicting the future. That purpose can best be fulfilled by using an approximation that is far more likely to be repeated than simple averaging of the 100 mutuel prices would indicate.

Using the list of sorted mutuel prices, pick the 25th from the bottom and the 25th from the top. Those values may be 8.00 and 4.00. Add to get 12.00. Divide by 2 to get 6.00--which is an accurate predictive value of what future average prices will be.

Multiply the 6.00 by 1.5 to get 9.00. Set that as a default value to replace any mutuel price greater than that amount and go through the list of mutuel prices again using:

If (mutuel price) > 9.00 then (mutuel price) = 9.00

Use the new cleaned data to calculate an ROI. If it is positive, you may have something that can usefully be applied to future races. Not guaranteed, but a much more dependable model than using dirty data.

BTW, the MIQR is used routinely in predictive computer modeling and simulations. It is not some personal quirk of mine.

Good and clear reply. Thanks.

[Capper Al]

Whatever method used it still all comes down to the wherewithal of the cappper. The computer won't make up for any personal handicapping deficiencies. The capper has to ask the right question. Many questions won't prove to work. Some work but aren't profitable. These are difficult to ignore. Matter of fact as a general rule, I keep them and wait for the public odds to stray.

[traynor]

I'm still unclear, why are you are using a factor of 1.5? How do you know that it represents a "reasonably accurate model of future potential"?

Thanks,

It is a standard calculation used in predictive computer modeling. I didn't select it arbitrarily.

[raybo]

Whatever method used it still all comes down to the wherewithal of the cappper. The computer won't make up for any personal handicapping deficiencies. The capper has to ask the right question. Many questions won't prove to work. Some work but aren't profitable. These are difficult to ignore. Matter of fact as a general rule, I keep them and wait for the public odds to stray.

Or wagering deficiencies! :lol: That's the killer for most players, no discipline, no patience, no consistency, underbetting or overbetting, chasing losses, etc., etc., etc.. I believe handicapping, once one has been around the game for a while, is the least of a players worries. When to bet, how to bet, and how much to bet is extremely important, IMO.

One's selections may be good enough, but which ones to wager on, and when to wager, how to wager and how much, is the kicker.

[Capper Al]

[QUOTE=raybo]Or wagering deficiencies! :lol: That's the killer for most players, no discipline, no patience, no consistency, underbetting or overbetting, chasing losses, etc., etc., etc.. I believe handicapping, once one has been around the game for a while, is the least of a players worries. When to bet, how to bet, and how much to bet is extremely important, IMO.

One's selections may be good enough, but which ones to wager on, and when to wager, how to wager and how much, is the kicker.[/QUOTE

Agree. The implementation of a system is a whole other layer of complexity, and that's wagering. Here again the capper is challenged as to their having the wherewithal or not. Computers won't and can't do it for you.

[raybo]

[QUOTE=raybo]Or wagering deficiencies! :lol: That's the killer for most players, no discipline, no patience, no consistency, underbetting or overbetting, chasing losses, etc., etc., etc.. I believe handicapping, once one has been around the game for a while, is the least of a players worries. When to bet, how to bet, and how much to bet is extremely important, IMO.

One's selections may be good enough, but which ones to wager on, and when to wager, how to wager and how much, is the kicker.[/QUOTE

Agree. The implementation of a system is a whole other layer of complexity, and that's wagering. Here again the capper is challenged as to their having the wherewithal or not. Computers won't and can't do it for you.

Well, some say their computers can do it for them, ie, automated betting. I wouldn't doubt that the Benters out there have that capability. But, for the rest of us, it's all up to us, we either sink or swim, and the computer is no life vest, we still have to do what it suggests that we do, or what's the use in having it in the first pace. That is of course, if your computer method does those kinds of things for you.

[OverlayHunter]

Say you have a sample of 100 races. If the mutuel prices are sorted from lowest to highest, the lowest may be 2.20 and the highest 89.00. The purpose of modeling is predicting the future. That purpose can best be fulfilled by using an approximation that is far more likely to be repeated than simple averaging of the 100 mutuel prices would indicate.

Using the list of sorted mutuel prices, pick the 25th from the bottom and the 25th from the top. Those values may be 8.00 and 4.00. Add to get 12.00. Divide by 2 to get 6.00--which is an accurate predictive value of what future average prices will be.


Traynor, I see the merit of your approach (thank you) and would like to explore it. I may be making this more complicated than it needs to be but I want to make sure I understand it and, as you will see from my logic below, I clearly don't understand it.

For the purpose of the clarification I'm seeking, let's assume you have a sample of 100 races and the set of factors you are testing produced 10% winners from the 100 races (which, of course, would be 10 winners) with a profitable ROI of 1.10.

In the first paragraph of the quote above, you speak in terms of "...a sample of 100 races" then "...averaging of the 100 mutuel prices would indicate." If you take the 100 mutuel prices from a 100 race sample surely some of those prices must be $0.00 as no set of factors ever produces 100% winners in any reasonable sample size and in my example of 10% winners you would have 90 such $0.00 mutuels and the 25th from the top and bottom would each be $0.00. Is that correct?

In your 2nd paragraph example, if the $8.00 was the 25th from the bottom, would that suggest the sample you are using as an example had at least 75% winners? If that is not the case, should the very first sentence of the first paragraph actually read "Say you have a sample of 100 winning races."

As I said, I'm not grasping this correctly.

[limit2]

Yes, I have a set of criteria for determining playable races. In fact to me it is primary that I consider the race before I consider the horses within the race. It goes something like this: I separate the races into 3 classes as: Open, Acceptable and Questionable by looking at the lowest ML in the program listing. Odds of 3:1, 5:2 and 2:1 aid in the determination.

[Dave Schwartz]

Agree. The implementation of a system is a whole other layer of complexity, and that's wagering. Here again the capper is challenged as to their having the wherewithal or not. Computers won't and can't do it for you.

As a player that uses nothing but the computer, with absolutely zero subjectivity on the part of the user, I would have to disagree.

What I described in the previous sentence, is the precise goal of almost every one of my software clients. Some have achieved a positive expectation with this approach.

The idea that everyone must do something a particular way in order to be profitable is simply not accurate, any more than all successful golfers must learn to play right-handed.


In my personal approach, the system selects contenders, decides if some of the "almost contenders" should be added, creates an order to prefer the horses as wagers, decides how many to actually bet (i.e. some are singles, others spread 3,4, or more horses to win), how much to bet, and then with the click of a button creates the actual wagering file, which is then uploaded to my ADW.

[raybo]

As a player that uses nothing but the computer, with absolutely zero subjectivity on the part of the user, I would have to disagree.

What I described in the previous sentence, is the precise goal of almost every one of my software clients. Some have achieved a positive expectation with this approach.

The idea that everyone must do something a particular way in order to be profitable is simply not accurate, any more than all successful golfers must learn to play right-handed.


In my personal approach, the system selects contenders, decides if some of the "almost contenders" should be added, creates an order to prefer the horses as wagers, decides how many to actually bet (i.e. some are singles, others spread 3,4, or more horses to win), how much to bet, and then with the click of a button creates the actual wagering file, which is then uploaded to my ADW.

Dave, that's what I was alluding to, a completely automated handicapping , and wagering, system. Many have handicapping systems that are automated, but the user is still required to decide on the bettability of the race, how much they should wager, what types of bets to wager, and they must decide if they trust what the system is telling them to do, and they must ultimately place the wagers manually. My emphasis was/is on the wagering side of the system. If one's wagering method is not automated there is still much room for the user to screw up an otherwise viable, and profitable system. Discipline, and consistency, in doing with your money what the system tells you to, is lacking in many players, and their ultimate downfall.

[traynor]

Traynor, I see the merit of your approach (thank you) and would like to explore it. I may be making this more complicated than it needs to be but I want to make sure I understand it and, as you will see from my logic below, I clearly don't understand it.

For the purpose of the clarification I'm seeking, let's assume you have a sample of 100 races and the set of factors you are testing produced 10% winners from the 100 races (which, of course, would be 10 winners) with a profitable ROI of 1.10.

In the first paragraph of the quote above, you speak in terms of "...a sample of 100 races" then "...averaging of the 100 mutuel prices would indicate." If you take the 100 mutuel prices from a 100 race sample surely some of those prices must be $0.00 as no set of factors ever produces 100% winners in any reasonable sample size and in my example of 10% winners you would have 90 such $0.00 mutuels and the 25th from the top and bottom would each be $0.00. Is that correct?

In your 2nd paragraph example, if the $8.00 was the 25th from the bottom, would that suggest the sample you are using as an example had at least 75% winners? If that is not the case, should the very first sentence of the first paragraph actually read "Say you have a sample of 100 winning races."

As I said, I'm not grasping this correctly.

Substitute whatever number of winning races you have for 100. The process is to make the winning mutuels more predictive--meaning that in a small sample, one or two "outliers" (unlikely to repeat) can distort the perceived ROI. That is, the process uses only winning race mutuels, not the mutuels of all races. I apologize for not clarifying that.

[Ray2000]

Use the new cleaned data to calculate an ROI. If it is positive, you may have something that can usefully be applied to future races. Not guaranteed, but a much more dependable model than using dirty data.




Do I get a refund on any mutuel payoff "removed" because it exceeded 1.5 x MIQR?

Just kidding traynor, :) ..The Mean of the InterQuartile Range is a good way to clean the results.

And as you mentioned bootstrapping or MonteCarlo can also be used to see if the sample data is representative.

[whodoyoulike]

Do I get a refund on any mutuel payoff "removed" because it exceeded 1.5 x MIQR?

Just kidding traynor, :) ..The Mean of the InterQuartile Range is a good way to clean the results.

And as you mentioned bootstrapping or MonteCarlo can also be used to see if the sample data is representative.

Is it really as he presented it? You can do a number of statistical calculations to attempt to demonstrate a point or idea. Whether it's valid, the way it is presented, is debatable. I'm familiar with calculating an average of the second and third quartiles and excluding the first and last quartiles. Comparing this calculation with your number provides a reference point. His method makes a comparison using an adjusted average of two values in the 2nd and 3rd quartiles and then goes through the entire process until all the values are used. If it's positive, your number may be useful. I'm still unclear why his formula utilizes a factor of 1.5 (why not 0.9 or 1.327?). I saw his response that the use of 1.5 is just part of the formula and not his selection. By using 1.5, you're comparing your number with an average that is adjusted towards the higher end of your list (why?). I frequently use averages in my calculations and comparisons. Should I make an adjustment of 1.5 for all of them?

I don't intend to offend anyone on this forum but, when someone presents an idea using a factor (he could have pulled out of his ass) and says it's a gold nugget because it has a yellowish color. I say it still looks like a piece of crap.

[Capper Al]

As a player that uses nothing but the computer, with absolutely zero subjectivity on the part of the user, I would have to disagree.

What I described in the previous sentence, is the precise goal of almost every one of my software clients. Some have achieved a positive expectation with this approach.

The idea that everyone must do something a particular way in order to be profitable is simply not accurate, any more than all successful golfers must learn to play right-handed.


In my personal approach, the system selects contenders, decides if some of the "almost contenders" should be added, creates an order to prefer the horses as wagers, decides how many to actually bet (i.e. some are singles, others spread 3,4, or more horses to win), how much to bet, and then with the click of a button creates the actual wagering file, which is then uploaded to my ADW.

Dave, From what I can tell you have a fine product. Using the box (purchasing software) or developing your own can lead a capper to the same end. If a capper studies and gets to know what they have purchased then they should be able to be profitable as if they developed the software themselves. Either way it's an effort.

[Capper Al]

Dave, that's what I was alluding to, a completely automated handicapping , and wagering, system. Many have handicapping systems that are automated, but the user is still required to decide on the bettability of the race, how much they should wager, what types of bets to wager, and they must decide if they trust what the system is telling them to do, and they must ultimately place the wagers manually. My emphasis was/is on the wagering side of the system. If one's wagering method is not automated there is still much room for the user to screw up an otherwise viable, and profitable system. Discipline, and consistency, in doing with your money what the system tells you to, is lacking in many players, and their ultimate downfall.

If one is not skeptical of the picks on occasion then they won't be profitable because they haven't a clue about the game. One still has to get to know the game.

[Overlay]

If one is not skeptical of the picks on occasion then they won't be profitable because they haven't a clue about the game. One still has to get to know the game.
I'm not sure that I understand your comment. If you know exactly why/how the pick and wagering strategy were arrived at, why would you ever be skeptical of them? Do you mean that there will always be isolated exceptions that a generalized or automated model won't be able to adequately or accurately assess?

[Capper Al]

I'm not sure that I understand your comment. If you know exactly why/how the pick and wagering strategy were arrived at, why would you ever be skeptical of them? Do you mean that there will always be isolated exceptions that a generalized or automated model won't be able to adequately or accurately assess?

Even if you were certain, there will be a few times your software reads garbage data in for a horse. I have had to call BRIS about their data a few times. Of course, I'll say that BRIS does a good job. It's only been a couple of times. Then one gets a race for Arabian horses in race 11 on Friday and has to make up data. It's an art.

I believe it might have been you who separates their races for playing or not. Always handicap every race and, at least, make a paper bet if not a real one. You'll be surprised what one can learn from races outside of their scope. I play at least 98% of all races and am profitable. I kid my racing partner to mark the calender when I skip a race. I think I'll be playing that Arabian race.

[traynor]

Is it really as he presented it? You can do a number of statistical calculations to attempt to demonstrate a point or idea. Whether it's valid, the way it is presented, is debatable. I'm familiar with calculating an average of the second and third quartiles and excluding the first and last quartiles. Comparing this calculation with your number provides a reference point. His method makes a comparison using an adjusted average of two values in the 2nd and 3rd quartiles and then goes through the entire process until all the values are used. If it's positive, your number may be useful. I'm still unclear why his formula utilizes a factor of 1.5 (why not 0.9 or 1.327?). I saw his response that the use of 1.5 is just part of the formula and not his selection. By using 1.5, you're comparing your number with an average that is adjusted towards the higher end of your list (why?). I frequently use averages in my calculations and comparisons. Should I make an adjustment of 1.5 for all of them?

I don't intend to offend anyone on this forum but, when someone presents an idea using a factor (he could have pulled out of his ass) and says it's a gold nugget because it has a yellowish color. I say it still looks like a piece of crap.

Use whatever number you feel like using, or that you believe will produce better results.

[whodoyoulike]

Use whatever number you feel like using, or that you believe will produce better results.

I was going let your response go as you stated that it was part of the formula. Then I noticed others might be tempted to use your method. Using your example, you have 100 values in numerical order. You eliminate values 1 through 25 and 76 through 100 which happen to be of various values. The remaining 50 values happens to be 100. The average should be 100. Your method would give you a value of 150 for comparison purposes. I couldn't see how that could be correct.

Again, I don't want to offend anyone. I'm was just curious, why the formula would use 1.5?

No response needed.

[DeltaLover]

I'm was just curious, why the formula would use 1.5?


There does not need to be an ansewer to why. The process of deriving a constant can be a trial and error process having it as an unkown to be optimized based in a specific fitness function. As bettors we are not concerned about analytical proves but for a working model.

[Overlay]

I believe it might have been you who separates their races for playing or not. Always handicap every race and, at least, make a paper bet if not a real one. You'll be surprised what one can learn from races outside of their scope. I play at least 98% of all races and am profitable. I kid my racing partner to mark the calender when I skip a race. I think I'll be playing that Arabian race.
I seldom treat a race as unplayable out-of-hand -- only on the very infrequent occasions when I'm missing (and also unable to reconstruct) some fundamental piece of data that I use, or when the entire field is so lacking in form that I can't project the outcome with an acceptable degree of confidence. Since I base my play on comparing my fair odds with tote odds/payoffs in order to determine which horses or combinations are being underbet, then the more races that are played, the greater the opportunity for those favorable percentages to assert themselves.

[raybo]

If one is not skeptical of the picks on occasion then they won't be profitable because they haven't a clue about the game. One still has to get to know the game.

What I meant by that was that occasionally your system might put you on a horse or horses that look absolutely pathetic and the odds reflect that, do you go against the system, or do you stick with it? That is where many players make big mistakes. Any system, whether automated or not, consists of the aggregate, the randomness included, the variance included. If one suddenly decides to turn perceptions into reality, and pass betting that horse, or worse, bet against that perceived randomness or perceived variance, one is in effect, destroying the system. It's like having a system that relies, to a large degree, on even odds or lower selections, and one suddenly decides to not bet those horses, then that system no longer exists. Because a portion of the aggregate is gone, so the system itself is no longer valid, regarding it's previous level of effectiveness.

Known bad data is one thing, but perceived bad data is just that, perceived, not known.

Reality is the collection of all individual events, each having it's own degree of influence, direct or indirect, on the overall outcome. Removing or altering any of those events, alters reality. Removing something from the past alters the present, and probably the future.

[Maximillion]

I dont think i could ever reach the point of removing all subjectivity from my game,(i.e. a system type approach) I need to take a hard look at the horses im betting money on....perhaps its a weakness.

I think ultimately it all comes down to price....If I was a "system" type player, I would absolutely remove all subjectivity if the system was pointing to long-odds horses.

[raybo]

I dont think i could ever reach the point of removing all subjectivity from my game,(i.e. a system type approach) I need to take a hard look at the horses im betting money on....perhaps its a weakness.

I think ultimately it all comes down to price....If I was a "system" type player, I would absolutely remove all subjectivity if the system was pointing to long-odds horses.

Mid to large priced horses is the backbone of my method, but mostly because of the tracks I play, those having higher than average win payouts. Anyone playing those tracks and constantly betting favorites or near favorites will drown, if they are playing a significant percentage of the races carded.

[traynor]

I was going let your response go as you stated that it was part of the formula. Then I noticed others might be tempted to use your method. Using your example, you have 100 values in numerical order. You eliminate values 1 through 25 and 76 through 100 which happen to be of various values. The remaining 50 values happens to be 100. The average should be 100. Your method would give you a value of 150 for comparison purposes. I couldn't see how that could be correct.

Again, I don't want to offend anyone. I'm was just curious, why the formula would use 1.5?

No response needed.

The value would not be 150 (unless your average mutuel is in the $100 range). It would be 1.5 times the average (or mean) of the middle half of the races. Add the values of the middle half, divide by the number of the races in the middle half, then multiply that average (or mean if you prefer) by 1.5 (or whatever).

The values are sorted from lowest to highest. The average (or mean) of the second and third quartiles (the middle half of the races) are far more likely to be representative of a larger sample. 1.5 is a common value used in data modeling, but it is not carved in stone. Just recommended. I used 2 for awhile (with all that standard deviation, sum of the squares, square root drivel mixed in) but found 1.5 worked better (for me). After a LOT of experimentation on endless chunks of races, I settled on the process described as being just as efficient, and only requires a couple of lines of code.

If I can code a few lines that do the same thing as a far more complex process--unless the complex process earns more profit for me--I will go with the few lines of code every time.

[Robert Fischer]

I dont think i could ever reach the point of removing all subjectivity from my game,(i.e. a system type approach) I need to take a hard look at the horses im betting money on....perhaps its a weakness.

I think ultimately it all comes down to price....If I was a "system" type player, I would absolutely remove all subjectivity if the system was pointing to long-odds horses.

I like the subjectivity as well. People can choose as well/or better than machines for the most part. Getting to the part where a choice is necessary is a pretty enormous job however, especially if we are talking all the tracks, all the horses. I'm lucky if I cover 3 tracks + a few spot play/watch listers.

I've thought about getting this stuff programmed.

I know some things that would be good to be programmed, and then it would be the best performance thing, at least from what we know about(it's not like we have access to Benter's and similar players' programs).
With a very conservative set of rules I would guess it could have a small number of auto play races that churn a fair ROI.
However, I would happily man it to access the vast majority of plays, until I got to the point where I wanted to free myself and train some others to competently use it.
(If I manned it, as opposed to both my current hands-on work, and/or a very conservative fully-automated system)It would certainly greatly increase the amount of races that I could have an advantage in, and greatly reduce the number of races that I miss :bang:.

As it stands, once in a while I don't reach my target goals and more than half of those days this month were due to lack of playable races. (If I had to guess, I would say I had at least 1 playable race on 75% of the days this month.)

It would be unwise to get a horseplayer to program this for me. He would likely use it and/or spread the knowledge. It is also too expensive for me to outsource at this time, and too expensive to obtain the necessary data at this time.

Maybe this will be the mark I leave on the game before I pass?

I don't know. It would be something I would have to consider the pros and cons, and what kind of effect it would have, should it become mainstream(not only for me personally, but if it made the parimutuel pools that much harder, I'm not sure that would be a good thing).

[OverlayHunter]

Traynor, thanks for the clarification and the insight. Greatly appreciated.

[Overlay]

It would be unwise to get a horseplayer to program this for me. He would likely use it and/or spread the knowledge.
Would it be possible to address that concern through the use of a full-field, fair-odds perspective, or does your handicapping not lend itself to being structured or expressed in that manner?

[Capper Al]

What I meant by that was that occasionally your system might put you on a horse or horses that look absolutely pathetic and the odds reflect that, do you go against the system, or do you stick with it? That is where many players make big mistakes. Any system, whether automated or not, consists of the aggregate, the randomness included, the variance included. If one suddenly decides to turn perceptions into reality, and pass betting that horse, or worse, bet against that perceived randomness or perceived variance, one is in effect, destroying the system. It's like having a system that relies, to a large degree, on even odds or lower selections, and one suddenly decides to not bet those horses, then that system no longer exists. Because a portion of the aggregate is gone, so the system itself is no longer valid, regarding it's previous level of effectiveness.

Known bad data is one thing, but perceived bad data is just that, perceived, not known.

Reality is the collection of all individual events, each having it's own degree of influence, direct or indirect, on the overall outcome. Removing or altering any of those events, alters reality. Removing something from the past alters the present, and probably the future.

I appreciate your logic. Simply put- if one hops of the system when they think the system may be wrong, then they are not playing their alleged system.

Some angles at times will jump out at me and scream for a wager. I collected $400 on a buck tri by jumping on a Euro running on an off turf. My system didn't pick this up, and I doubt most systems would. For me, it's man and machine. But then I don't have a system as nice as Dave's that ties right into the tote-board either.

[raybo]

I appreciate your logic. Simply put- if one hops of the system when they think the system may be wrong, then they are not playing their alleged system.

Some angles at times will jump out at me and scream for a wager. I collected $400 on a buck tri by jumping on a Euro running on an off turf. My system didn't pick this up, and I doubt most systems would. For me, it's man and machine. But then I don't have a system as nice as Dave's that ties right into the tote-board either.

I'm sure one might seemingly capitalize by occasionally going against their method, but are they sure those instances reflected an overall gain? You stated a single $400 gain, but did all your betting irregularities, regarding the system, translate into a net gain or net loss? That's the key, IMO. In other words, how much did your "hunches" make and how much did they lose?

[Capper Al]

I'm sure one might seemingly capitalize by occasionally going against their method, but are they sure those instances reflected an overall gain? You stated a single $400 gain, but did all your betting irregularities, regarding the system, translate into a net gain or net loss? That's the key, IMO. In other words, how much did your "hunches" make and how much did they lose?

All I can say is that is a good question. My racing partner keeps track of that stat and claims to be ahead by jumping off.

[raybo]

All I can say is that is a good question. My racing partner keeps track of that stat and claims to be ahead by jumping off.

Well, as long as he is tracking all those events, then good for him, go for it!

[Robert Fischer]

Would it be possible to address that concern through the use of a full-field, fair-odds perspective, or does your handicapping not lend itself to being structured or expressed in that manner?
I am interested Overlay. Here is my answer, and I may need more input from you, if I am not addressing exactly what you are asking.

My worry is that with the value being somewhat unique(I'm not reinventing the wheel here, but it is a unique way of performance rating, and (a large)part of it's usefulness lies in the fact that the process uncovers some things that mainstream performance indicators do not, and in order to calculate them, they would normally take a manual hands on slow approach from an experienced handicapper.

This isn't the Sheets, but what I'm talking about could probably be lumped in with that same genre. Stripped away from the specifics, it takes trip into effect and measures things that is not measured by other performance standards(including sheets/graphs/etc..).

ODDSLINE??? Now my personal approach is favorite-centric. Part of this is the time consuming nature because I have to manually do this work, and thus I go for the heart of the matter. More on Favorite Centric approach= No! lol I'm not a chalk eating weasel, but what I am saying, is that it's my theory that with the parimutuel system the favorite is very important to have a beat on. My favorite is play-against favorites(because there is then so much money in the pools left over), but strong favs are great as well(the high hit% they bring along with that grind out approach is what stabilizes a player who isn't funded buy Bill Gates, and reduces the risk from somewhere near "Gambling" to somewhere near "Investing", or Speculating, or whatever kinder gentler word. What makes Mr. Fischer an unhappy man, is races I don't know(PASS), or races where the favorite can not be bet against nor on(pass).

Lend itself to being structured or expressed in the manner of a full-field, fair-odds perspective??

yes.
Part of the advantage of automating this workload would be opening new doors when it comes to moving beyond a limited favorite centric approach that manual calculation calls for. Multi-horse opportunities like vertical exotics(in spite of takeout, and the multi-race craze, these bets still provide some of the best spot-plays in the game), Dutching, Longshot value plays, and even post race things like new perspectives on traditional looks like Key Race analysis, all come to mind as things a favorite centric player like myself would want to play around with given the chance. Probably missing some as well.

[Overlay]

I am interested Overlay. Here is my answer, and I may need more input from you, if I am not addressing exactly what you are asking.

My worry is that with the value being somewhat unique(I'm not reinventing the wheel here, but it is a unique way of performance rating, and (a large)part of it's usefulness lies in the fact that the process uncovers some things that mainstream performance indicators do not, and in order to calculate them, they would normally take a manual hands on slow approach from an experienced handicapper.

This isn't the Sheets, but what I'm talking about could probably be lumped in with that same genre. Stripped away from the specifics, it takes trip into effect and measures things that is not measured by other performance standards(including sheets/graphs/etc..).

ODDSLINE??? Now my personal approach is favorite-centric. Part of this is the time consuming nature because I have to manually do this work, and thus I go for the heart of the matter. More on Favorite Centric approach= No! lol I'm not a chalk eating weasel, but what I am saying, is that it's my theory that with the parimutuel system the favorite is very important to have a beat on. My favorite is play-against favorites(because there is then so much money in the pools left over), but strong favs are great as well(the high hit% they bring along with that grind out approach is what stabilizes a player who isn't funded buy Bill Gates, and reduces the risk from somewhere near "Gambling" to somewhere near "Investing", or Speculating, or whatever kinder gentler word. What makes Mr. Fischer an unhappy man, is races I don't know(PASS), or races where the favorite can not be bet against nor on(pass).

Lend itself to being structured or expressed in the manner of a full-field, fair-odds perspective??

yes.
Part of the advantage of automating this workload would be opening new doors when it comes to moving beyond a limited favorite centric approach that manual calculation calls for. Multi-horse opportunities like vertical exotics(in spite of takeout, and the multi-race craze, these bets still provide some of the best spot-plays in the game), Dutching, Longshot value plays, and even post race things like new perspectives on traditional looks like Key Race analysis, all come to mind as things a favorite centric player like myself would want to play around with given the chance. Probably missing some as well.

My point was more along the lines of addressing your concerns about the dissemination of a method of play, which carries with it the very real possibility of driving down mutuels below the point of profitability (assuming that the method operates on the principle of narrowing a field to a single selection -- whether it be the favorite or some other horse in the field -- that would be the same for anybody using the method). I was wondering whether it would be more helpful to base selections upon development of a full-field fair-odds line that could point up wagering value wherever it might exist, and that would not limit consideration to just the single horse isolated by any particular method, or that at least would tell you whether that single horse was worth a bet at its posted odds.

[Robert Fischer]

My point was more along the lines of addressing your concerns about the dissemination of a method of play, which carries with it the very real possibility of driving down mutuels below the point of profitability (assuming that the method operates on the principle of narrowing a field to a single selection -- whether it be the favorite or some other horse in the field -- that would be the same for anybody using the method). I was wondering whether it would be more helpful to base selections upon development of a full-field fair-odds line that could point up wagering value wherever it might exist, and that would not limit consideration to just the single horse isolated by any particular method, or that at least would tell you whether that single horse was worth a bet at its posted odds.

A full-field fair-odds line is definitely missing from my current game, and could help a lot(especially if I took things a step further to automate some of the process and free up resources,time,etc).

[WORKFORCE]

Any ideas on how to make your own class pars based on times rather than purse valuation?. At the moment, I have universal standards at each distance along with additional racecourse standards at each class level but I'm struggling to convert the data into a realistic class hierarchy.

What I've done is to try working out the difference in seconds from the racecourse standards from the universal standards for example; Lingfield's 5f standard is 5 seconds quicker than the Universal (All course average) and if you divide that by .17s per length then multiply by 3.7lbs per length you start getting silly figures.

[Dave Schwartz]

Any ideas on how to make your own class pars based on times rather than purse valuation?. At the moment, I have universal standards at each distance along with additional racecourse standards at each class level but I'm struggling to convert the data into a realistic class hierarchy.

1. Begin with a numeric class system that quantifies the "level" of each race into something that (at least aspires to) represents lengths between finishers.

2. Use these class levels to group winning horses together in similar class groups for each distance run.

3. Take the races from each group and do some mathematical razz-a-ma-tazz to come up with a rating. (Example: toss out the fastest and slowest 10% and average the rest.)

This will give you a set of par times for each level and distance.

4. Use these resultant "level-based pars" to compare the projected winning times at the track with the actual winning times.

5. Begin predicting race winners based upon a systematic approach of utilizing the resultant speed figures. Track the "success metric." For us, it is simply the percentage of races where one of the top 3 horses in the race wins. More specifically, we use the best-of-last-2 speed ratings in sprint races and "best-2-of-last-3" in route races.

In the U.S. the metric has been consistently 62% for over 20 years. Amazingly, field size seems to have very little impact.

6. Feed this metric back into your system for improvement. That is, ask the question, "does this track produce results within an acceptable range?"

7. Compare the projected winning ratings for each class level to the actual winning ratings. Are they out of whack?

8. Return to #1 after you have adjusted your class levels. It is a continual process of:


Improve class levels
Build speed ratings
Compare to "success metric."
Repeat


Hope this helps.


Regards,
Dave Schwartz

PS: Here is a document that I wrote in 1991 and updated about 3 years ago that might be of help:

http://www.practicalhandicapping.com/desktop/package/External/2010Pars/

[Capper Al]

Any ideas on how to make your own class pars based on times rather than purse valuation?. At the moment, I have universal standards at each distance along with additional racecourse standards at each class level but I'm struggling to convert the data into a realistic class hierarchy.

What I've done is to try working out the difference in seconds from the racecourse standards from the universal standards for example; Lingfield's 5f standard is 5 seconds quicker than the Universal (All course average) and if you divide that by .17s per length then multiply by 3.7lbs per length you start getting silly figures.

Wow! You are working with European racing? If we wanted to do what you want to here in North America, we would use par times as the standard from our data suppliers such as BRIS. Understanding class from speed is a debatable topic. I side with purse valuation. Let's start a thread on this in the Capper's Corner.

[raybo]

Wow! You are working with European racing? If we wanted to do what you want to here in North America, we would use par times as the standard from our data suppliers such as BRIS. Understanding class from speed is a debatable topic. I side with purse valuation. Let's start a thread on this in the Capper's Corner.

I think we have tossed class back and forth quite a bit in the recent, and distant past, and to the best of my recollection class has not even been universally defined. If you have hopes of getting anywhere constructive shouldn't there be at least a basic understanding that everyone can agree on regarding what "class" is? Otherwise how can any progress be made on this subject of class pars? I certainly would be interested in collaborating but not until the majority of potential contributors agrees, generally, on what we're talking about when we mention class. As a precursor, I personally will not entertain any idea that the definition of class contain any inclusion of man-made class levels, or straight purse levels. Both of those can be so far from anything meaningful that their inclusion is contra to anything of value, IMO.

[WORKFORCE]

1. Begin with a numeric class system that quantifies the "level" of each race into something that (at least aspires to) represents lengths between finishers.

2. Use these class levels to group winning horses together in similar class groups for each distance run.

3. Take the races from each group and do some mathematical razz-a-ma-tazz to come up with a rating. (Example: toss out the fastest and slowest 10% and average the rest.)

This will give you a set of par times for each level and distance.

4. Use these resultant "level-based pars" to compare the projected winning times at the track with the actual winning times.

5. Begin predicting race winners based upon a systematic approach of utilizing the resultant speed figures. Track the "success metric." For us, it is simply the percentage of races where one of the top 3 horses in the race wins. More specifically, we use the best-of-last-2 speed ratings in sprint races and "best-2-of-last-3" in route races.

In the U.S. the metric has been consistently 62% for over 20 years. Amazingly, field size seems to have very little impact.

6. Feed this metric back into your system for improvement. That is, ask the question, "does this track produce results within an acceptable range?"

7. Compare the projected winning ratings for each class level to the actual winning ratings. Are they out of whack?

8. Return to #1 after you have adjusted your class levels. It is a continual process of:


Improve class levels
Build speed ratings
Compare to "success metric."
Repeat


Hope this helps.


Regards,
Dave Schwartz

PS: Here is a document that I wrote in 1991 and updated about 3 years ago that might be of help:

http://www.practicalhandicapping.com/desktop/package/External/2010Pars/

Thank you so much for the post, that goes to Capper Al & Raybo also.

I have signed on just to let you know that I appreciate the response, in case you was wondering where I went to regarding this topic. What I will try to do is digest your points 1-8 along with the document you've sent and I will hopefully get back to you with a more meaningful reply once I've integrated a few ideas into the system where I might be able to pose some more thoughts on the subject.

Workforce.

[Dave Schwartz]

You are welcome.