Best betting strategy for grinding it out [Archive] - Horse Racing Forum - PaceAdvantage.Com

Dave Schwartz

06-06-2006, 06:21 PM

You should market this HMI. It's awesome!

Amazingly, we have sold over 10,000 copies of this since its release in 1990.

I mean that Mo Ali guy sells a system that is basically as old as time but yours seems to be incredible. Nice job there dude. Seriously you are the man. I'm going over to your site and dig in. I have a feeling there's a lot of good stuff there. And yes I do spend dinero on what I feel is good stuff....thanks

We have operators standing by. <G> That number in New Jersey is Klondike 5...

Seriously, thank you for the kind words. I await both your order and your feedback.

Dave

PS: That Ali stuff appears to be a bit weak, doesn't it?

06-06-2006, 06:50 PM

Since most of your profit is rebate, the key is to churn as much money as possible through the windows. Find a happy medium between bet size and volume of races played. The more races, the less your bets will influence your payout. If you bet lots of races at a smaller rate, you can keep the smaller tracks on your betting menu. If you bet a few races, you will have to stick to the big pools to make money you want.

Learn the pool sizes at the tracks you will play, and the most efficient wager size for that track. This can vary according to day of the week, field size, class, track condition, surface, etc. This may be substantially below what you can comfortably bet according to your bankroll, but will maximize profits. It is possible to overbet the place pools to the point of turning a paper profit into a loss. Don't expect to generate the same overall profit betting place at Thistledown or Calder as you will at Belmont.

Side note, if you think your money isn't going into the pools and you are consistently winning, I'd think again.

None of this is a knock on Dave's HMI, which I bought many years ago and think is great.

westbridge

06-06-2006, 10:53 PM

I'd love to. <G>

http://www.horsestreet.com/products/hmi/index.html

It might not be fair to imply something is wrong with Kelly using a ROI test.

Kelly wasn't meant to maximize ROI. It was to maximize wealth growth, i.e. Final Wealth/Initial Wealth. Kelly did its job in your examples.

Dave Schwartz

06-07-2006, 01:12 AM

West,

It might not be fair to imply something is wrong with Kelly using a ROI test.

Kelly wasn't meant to maximize ROI. It was to maximize wealth growth, i.e. Final Wealth/Initial Wealth. Kelly did its job in your examples.

Can't disagree with that statement. However, I contend that the majority of players simply cannot afford any strategy that lowers their ROI.

Also, the entire concept of Kelly is based upon the accuracy of probability and odds. Couple the fact that the true probability of an individual horse is almost never "accurate" and the ever-changing odds and you have a concept that simply does not work in the real world.

Understand that I am not saying that a player cannot win using Kelly. I am saying that the concept simply does not function accurately the way most players use it.

I truly believe that any player who can get a "reasonably high" hit rate and even a small profit margin of 3% or so can make a living at the races. I have seen it done and I have done it myself.

Regards,
Dave Schwartz

spilparc

06-07-2006, 01:24 AM

West,

Understand that I am not saying that a player cannot win using Kelly. I am saying that the concept simply does not function accurately the way most players use it

I've always believed that Kelly betting risked too much of one's bankroll. It looks good on paper, but in the real world it just doesn't quite work out the way most people think it will.

Dick Mitchell proposed a 1/2 Kelly, but I think that is too risky for the average investor.

Dick Schmidt

06-07-2006, 02:19 AM

"It might not be fair to imply something is wrong with Kelly using a ROI test.

Kelly wasn't meant to maximize ROI. It was to maximize wealth growth, i.e. Final Wealth/Initial Wealth. Kelly did its job in your examples."

Actually, Kelly was meant to maximize the efficiency of telephone switching equipment. As far as anyone knows, Kelly never saw a horse race or made a bet in his life. He used racing as an example to simplify how his idea worked and it took on a life of its own.

I've known lots of professional gamblers, both racetrack and day traders, and none of them use Kelly, 1/2 Kelly or anything that moves around that much. I did meet one trader who took 3 1/2 million bucks to $750,000 in four days trying Kelly. It can rip you up some. Be careful out there.

Dick

If someone with multiple personalities threatens to kill himself, is it considered a hostage situation?

Topcat

06-07-2006, 02:33 AM

That leads me to a question:

Dick, Do you use any money management strategy in commodity trading?

I actually tried 1/4 Kelly and even HMI to a series of trades (thankfully on paper) and it made a good picture bad

Besides Trading in the Zone any recommended works/books/software on commodity trading?

hdcper

06-07-2006, 02:35 AM

So Dave, based on your article you wagered $535 and made a $98 profit. Why not just bet $25 flat per race and end up with a $100 profit (20% rather than $18) and risk less invested dollars ($500)?

Bill

westbridge

06-07-2006, 08:32 AM

The way I read it, Kelly's algorithm IS explicitly for maximizing capital growth (horse racing or other gamblering.)

http://www.bjmath.com/bjmath/kelly/kelly.pdf

Page 3 & 4. G = "exponential rate of growth of the gambler's capital" and the algorithm is the maximize G.

Anyway, my initial comment was not meant to criticize HMI. I was just to point out the difference in purpose. Perhaps it's a gambler's decision to maximize ROI or capital growth. And Dave pointed out rightly that Kelly requires very accurate probabilities and odds.

"It might not be fair to imply something is wrong with Kelly using a ROI test.

Kelly wasn't meant to maximize ROI. It was to maximize wealth growth, i.e. Final Wealth/Initial Wealth. Kelly did its job in your examples."

Actually, Kelly was meant to maximize the efficiency of telephone switching equipment. As far as anyone knows, Kelly never saw a horse race or made a bet in his life. He used racing as an example to simplify how his idea worked and it took on a life of its own.

I've known lots of professional gamblers, both racetrack and day traders, and none of them use Kelly, 1/2 Kelly or anything that moves around that much. I did meet one trader who took 3 1/2 million bucks to $750,000 in four days trying Kelly. It can rip you up some. Be careful out there.

Dick

If someone with multiple personalities threatens to kill himself, is it considered a hostage situation?

Dave Schwartz

06-07-2006, 12:52 PM

Hdcpr,

This was a short sample for illustration purposes. In a larger sample HMI will generally outperform both Kelly and flat betting.

Flat betting provides for very little growth, especially as the actual ROI approaches break-even.

As an example, I just ran a 60-race flat bet example with 20 winners at precisely 2/1. In that example the flat bettor shows no growth (of course), while the HMI user actually has a 2% positive ROI and a 9% gowth on bankroll.

Now, 2% may not seem like mouch but to be winner at all in a break-even situation is certainly, as they say, better than a sharp stick in the eye.

The strength of HMI is that in a slightly positive session (or even slightly negative if the session is long enough) it manages to show both an ROI improvement (over Kelly and flat) and positive growth. Flat betting cannot do that because it is precisely tied to flat profit.

As I state in the manuscript itself, HMI is not a something-for-nothing plan. If you are a losing player, using HMI will not cure that. I guess I could say that HMI is as close as one can get to "free lunch" in the wagering world.

Still, it does not come without trade-offs. When you get a tremendous rush and win lots of money, HMI's conservativeness will dial that back a notch or two. The whole idea is that most winning players experience a roller coaster ride, from peak to valley and back to peak, passing through the break-even point (zero line) many times over a sample of wagers. When you measure HMI from one zero line to the next, where the flat bettor is coming back to even (some here call that "regression to the mean"), HMI has gorund out sonsistent profit.

Personally, I am willing to trade some of the high side for improvement on the low side.

Dave

ryesteve

06-07-2006, 01:07 PM

The strength of HMI is that in a slightly positive session (or even slightly negative if the session is long enough) it manages to show both an ROI improvement (over Kelly and flat) and positive growth.

Wait a second... doesn't everyone concede that no money management system will turn a negative ROI into a positive ROI? But you're saying that "positive growth" can be attained in a "slightly negative" session, especially if it's a long one?? What am I missing here?

Dave Schwartz

06-07-2006, 01:46 PM

Wait a second... doesn't everyone concede that no money management system will turn a negative ROI into a positive ROI? But you're saying that "positive growth" can be attained in a "slightly negative" session, especially if it's a long one?? What am I missing here?

I am not suggesting that the long term result will be positive. I am saying that often the result will be positive.

Again, I reiterate (from the web page):

So what promises can I make? First of all I promise you that this is not a "something-for-nothing" approach to wagering. If you are a player that loses consistently because you are incapable of showing a flat-bet profit, I promise you that you will continue to lose. If you are not willing to begin keeping records of your wagers then you probably lack the commitment necessary for wagering success. If you expect this strategy to overcome an unprofitable selection method, you will be disappointed because no money management strategy can do that. Now the good news. If you are capable of showing even a marginal flat-bet profit, Horse Market Investing will make you tough to beat. Very tough. If you are capable of showing a strong, consistent advantage, HMI will make you virtually unbeatable.

Dave

ryesteve

06-07-2006, 02:00 PM

I am not suggesting that the long term result will be positive.
Kinda sounded like you were, especially when you said "if the session is long enough". Thanks for clarifying.

Barnstormer

06-07-2006, 03:42 PM

You have to "build a market". Only play those races in which you have confidence in your ability to predict true odds and only when the board odds are better. Even then you have to hope that late swings don't kill your advantage.

Easier sid than done, though, so I'd imagine that the majority of successful grinders stick with particular types of races that exhibit particular angles.

Off topic question: I'm trying to build a pars table for a few tracks and I'm confused by the nomenclature for race conditions. Say there's a claimer at Belmont with a purse of 44K and a claiming price of 25K. For class identification purposes, is that a 44K claimer or a 25K claimer?

PaceAdvantage

06-07-2006, 05:44 PM

25k claimer

andicap

06-09-2006, 03:47 AM

Dave,
could you email me so we could make arrangements for my split of your sales from this thread?

P.S. I don't think anyone is on to us.

;)

Dave Schwartz

06-09-2006, 09:58 AM

LOL - HMI is on backorder and should begin shipping again on MOnday. (All thanks to Andy.)

Now if I could just get you to do that with HSH. <G>

hdcper

06-11-2006, 01:59 AM

Dave,

Decided to take a look at your HMI manual and the example in the manual shows an example of a player wagering on 15 even money favorites (winning 8 of the 15 races). Anyway based on the pattern of wins and losses offered your HMI performance did outperform flat bet wagering (9.86% vs 6.67%).

However as I am sure you are aware, the win and loss pattern is the cause for this performance not the HMI method per say. Basically what I am trying to say, is if we look at a different win pattern and honestly I just picked a simple pattern at random (W,L,W,L and so on, the performance is pretty much just the opposite) as is reflected below(HMI 1.84% vs 6.67%):

HMI Player
Bet # Bet Result Value
1 50 50 1050
2 55 -55 995
3 50 50 1045
4 55 -55 990
5 49 49 1039
6 54 -54 985
7 49 49 1034
8 53 -53 981
9 48 48 1029
10 53 -53 976
11 48 48 1024
12 52 -52 972
13 47 47 1019
14 52 -52 967
14 47 47 1014
762 14

1.84%

Although HMI maybe an improved wagering technique, I suggest anyone looking at this method, consider doing many trial runs to determine when HMI really does outperform basic flat betting.

Hdcper

Dave Schwartz

06-11-2006, 02:22 AM

Flat betting stunts your growth.

Kelly stunts your ROI.

HMI gives you most of the advantages of Kelly without the disadvantage.

I have heard a few truly wonderful success stories with it over the years. One was an old guy from Louisiana who called me and told me that he was an old Sartin player, on social security, grinding out $200 each month with his small bankroll. After 3 months of using HMI, he said that his expectation was almost $3,00 per month.

Another guy, a highly structured win bettor, had 18 months of wagers in an Excel spreadsheet. In that sample of real bets he had made like 7,000 bets and his bankroll had grown by only $500. (I don't recall what he started with but I am thinking like $1,000 or so.) When he ran the same sample through HMI he was almost $8,000 ahead.

The best one of all is the basketball guys I used to write software for back in the '80s. (Note that I actually invented HMI in the '70s - and used it to make my living at blackjack. It was called Sports Market Investing then but not actually written yet.)

These guys started the 1988 NBA season with $30k, split evenly between a "Kelly-type" approach and HMI. A typical day for them was 4 bets and, as I recall, they might wager as much as 15-20% between the 4 plays.

Now, every year I would get a Christmas bonus from these guys but not this time. They were just to angry at me for suggesting HMI.

See, they were cruising along winning like 68% of their bets. The $15k in the Kelly BR had grown to about $160,000 while the HMI side had only (as I recall) grown to $38,000. "You cost us over a $120,000!" was what I heard about my bonus.

My response was to tell them that no sports handicapper in the history of basketball wins two-out-of-three with 4 bets per day and that they better get ready for a crash 'cause it's coming.

Well, the crash started just after New Years and continued right through the end of the season. They had two our three good spurts, one around the All-Star break and another just before the end of the season, but basically, they got their collective asses kicked the rest of the way.

See, while HMI has a tendency to be over-cautious in a big winning streak, it makes up for it by giving money back very slowly during losing streaks.

While the Kelly bankroll went into this huge tailspin and blew back all but $500 of the $145,000 it was ahead, HMI only surrendered back $6k from its high point. Final score was HMI $32,000 (plus $17k) and Kelly $15,500 (plus $500).

While this is just an anecdotal illustration that proves nothing, it is precisely how HMI works.

Again, however, I warn that one must have a real advantage or your will still lose (perhaps even faster than with Kelly).

Regards,
Dave Schwartz

Dave Schwartz

06-11-2006, 02:28 AM

One last thing...

The pattern thing is not as important as you think. It appears to be important in a small sample but I assure you that as the sample size increases the patterns become less important.

The exception to that is the very extreme, "unnatural" patterns.

By "unnatural" I mean patterns like 12 straight wins or 40 straight losses. Yes, I know they happen from time to time but they should be aberrational to your style of play. If they aren't then HMI is probably not a good match for your style of play.

Simply put, after using HMI for a significant session of play, you should see that your larger wagers have a higher number of winners than you smaller wagers. I think it is safe to say that this is what we would all like to strive for.

Dave

formula_2002

06-11-2006, 09:04 AM

I'd love to. <G>

http://www.horsestreet.com/products/hmi/index.html
Dave, the HMI table indicates a profit of 18%.
All your winners were 2-1. If I assume your losers were also 2-1, a dutch play would have returned a 28% profit.
Now if your bet size is determined by the final odds, its a fair comparison, else there is a bit of disadvantage to the dutch player,since the dutch is based on the final odds, and in many races, that may not be known to the bettor until the horses are at the 1/4 pole!! But that bit of disadvantage may be compensated for by the large difference in the roi.

20plays at 2-1, e= 20*.33=6.6
actual winners = 8
roi= a/e
roi = 1.275

Joe M

Dave Schwartz

06-11-2006, 10:40 AM

Difficult to dutch across multiple races.

BTW, in the previous anecdotal story of the "guy from louisian" I did not mean to infer that he was wagering the same amount of money. I believe that the new wagering approach gave him the confidence to increase his wager size.

Dave

formula_2002

06-11-2006, 11:59 AM

Difficult to dutch across multiple races.

Dave

Dave, give me an example of what you mean?
As long as I know the odds, I know the bet size.
The play is not bankroll dependent, but there does come a time, one may want to increase the play.
That is, instead of betting $33 on a 2-1 and getting back $100, one may want to up all the bets to $66 and then get back $200.

Dave Schwartz

06-11-2006, 03:29 PM

(I actually posted this answer in the wrong thread.)

Formula,

The following is meant to be constructive rather than critical in nature:

Dutch betting adjusts the size of the bet to the booking percentage

A 2-1 horse has a book % of .333 (1/(odds+1)

What you are referring to is not commonly called "dutching."

Dutching is where all the horses in a simultaneous wager (generally in the same race) are wagered on so that whichever horse wins the same amount is returned.

What you are reffering to is usually called "unit payoff."

This refers to wagering according to the booking percentages of the horses such that a 2/1 horse is wagered at 33 units, a 1/2 horse at 67 units, 4/1 at 20 units, etc.

If there is a skewing towards low-priced winners in a given strategy, this will improve your expectancy. Most losing players would benefit by adopting this strategy because they will wager more money on the lower-priced horses; something that, generally speaking, improves your net loss.

It also serves an excellent puprose when it is used in conjunction with flat wagers to tell whether the particular sample's $net was effected by an inordinate number of longshots.

Regards,
Dave Schwartz

formula_2002

06-11-2006, 09:24 PM

(I actually posted this answer in the wrong thread.)

Formula,

The following is meant to be constructive rather than critical in nature:

What you are referring to is not commonly called "dutching."

Dutching is where all the horses in a simultaneous wager (generally in the same race) are wagered on so that whichever horse wins the same amount is returned.

What you are reffering to is usually called "unit payoff."

This refers to wagering according to the booking percentages of the horses such that a 2/1 horse is wagered at 33 units, a 1/2 horse at 67 units, 4/1 at 20 units, etc.

If there is a skewing towards low-priced winners in a given strategy, this will improve your expectancy. Most losing players would benefit by adopting this strategy because they will wager more money on the lower-priced horses; something that, generally speaking, improves your net loss.

It also serves an excellent puprose when it is used in conjunction with flat wagers to tell whether the particular sample's $net was effected by an inordinate number of longshots.

Regards,
Dave Schwartz
Dave, that was not the "example" I hoping for was looking for.

Call it what you will, it seems to beat the HMI. Right?

formula_2002

06-12-2006, 06:31 AM

Perhaps you can show me your HMI results for the following results I recorded from "live selections for saturday", picks as posted on the selection forum?

26 plays, 8 winners, sum(1/(odds+)=6.46

a/e= 8/6.46= 1.238=roi

odds booking%
1.5 0.40
2.05 0.33
1.9w 0.34
1.3 0.43
5.2 0.16
1.3w 0.43
6.4 0.14
4.1w 0.20
8.9 0.10
3.5w 0.22
4.3 0.19
4.7 0.18
5.4 0.16
6.1 0.14
3.7w 0.21
3.2 0.24
2.6 0.28
5.7 0.15
1.1 0.48
2.9 0.26
0.8w 0.56
19.5w 0.05
2.9 0.26
1.7 0.37
12.2 0.08
6.9w 0.13
6.46

As far as the definition of the term "Dutching" your are correct. According to "Ainslie's Complete Guide To Thoroughbred Racing"(CR 1968),
page 61, the term is honor of it's supposed origionator, a player known as Dutch.
I must admit that very early, on I seized it for it's greater value, part of which you have already described. But Im off point, the more interesting question is, "What are the HMI results to the above table of picks?"

06-12-2006, 06:39 AM

formula_2002

06-12-2006, 06:54 AM

I can tell you one way to improve that series, and it is something that could help almost any win better. Stop betting the chalky ones! Stick to 3.5 and up, or even higher, it will almost always improve your results.

Incremental odds analysis, a topic I'm often writing about on the board, is imperative, and helps clarify what you are implying, but that's not what I'm talking about here. I simply want to know how HMI would play these picks.

Dave Schwartz

06-12-2006, 09:55 AM

Here is the output from an HMI spreadsheet:

I used the standard settings, with no tinkering. Obviously, a custom fit would likely do a better job.

In the example, a player starting with a $300 bankroll would have finished at $723.80, and had a $net of $4.02. That is, his overall advantage would have been 101%.

Hope this answers your question.

Dave

http://www.horsestreet.com/BBSImages/HMITest01.jpg

formula_2002

06-12-2006, 10:58 AM

banacek

06-12-2006, 11:23 AM

Dagve,

I'm having trouble understanding your spreadsheet. After 2 bets your bank is $122, then you win a $13 bet at 1.90-1 for a profit of $24.70 as stated, but your bank is now $159.70? Looks like the whole amount collected is there. Next win $13 at 1.30-1 and your bank goes up by $25.90 even thought your profit is only $16.90.

Or does BR mean soemthing else than bankroll - or is it too early in the morning for me?

ryesteve

06-12-2006, 11:34 AM

Yes, I was just about comment on the same thing. The bankroll column is getting incremented by the return on the bet, not profit. That can't be right...

Dave Schwartz

06-12-2006, 12:23 PM

Try again.

BR column is getting the "result" added to it.

Red Knave

06-12-2006, 12:36 PM

Yes, I was just about comment on the same thing. The bankroll column is getting incremented by the return on the bet, not profit. That can't be right...
What you and banacek are missing is the 'Buy' from the Reserve which is added to the Bankroll before the next wager. In the example banacek mentions, the 'Buy' and the wager are the same amount. The math goes like this -
BR + 'Buy' - Wager + Return ... so
$122 + 13 - 13 + 37.70 = $159.70

HTH

Dave Schwartz

06-12-2006, 12:54 PM

The column you should be looking at for the proper math is "BR."

The BR + Reserve = the total bankroll

formula_2002

06-12-2006, 01:27 PM

There is something you have to remember here with HMI or any other incremental betting strategy.
At the end the day (a very long day), if you are making 10% in every single odds range you cannot increase your % profit with any money management system
You cannot avoid a/e.

Bet 100 horses at $10 and 100 horses at $11, it’s still going to work out to 10%.

Everything is incremental. :bang:

Is there anyone that does not believe this?

ryesteve

06-12-2006, 01:28 PM

Try again.

BR column is getting the "result" added to it.
I guess it depends which BR column you were looking at... Knave's explanation was helpful. Without knowing the premise behind this method, it wasn't immediately apparent where the money was coming from.

Barnstormer

06-12-2006, 03:35 PM

I hit upon a neat trick that may or may not be better than a fractional Kelly strategy. I set up a spreadsheet to calculate optimal bets for win place and show. It simulates the pool payoff mechanisms and lets me input my best guess at the probabilities for all finish orders (how I come up with that is another topic entirely).

Then I set up a grid where I calculate the log of bankroll growth for each possible finish order. This is a LOT of formulas, so I create them using macros.
The sum of this big array of cells is the theoretical exponential rate of growth for a given spread of bets.

Thirdly, I use the Solver to maximize this growth factor by varying the bets. I'll warn anyone who wants to try this NOT to use integer bet sizes, since the Solver will usually generate zero bets or else not find the optimum.

Finally, and here's the cool part: I then run another Solver run, this time to find the SMALLEST spread of bets that will generate 90% of the maximum theoretical growth factor. This is an ad hoc way of finding an acceptably small kelly fraction.

In trials I've found that it often produces total bet sizes one quarter or less than the full Kelly bet spread while maintaining most of the advantage.

I haven't tried this method for real money yet, but it looks promising.

Dave Schwartz

06-12-2006, 05:16 PM

Now I understand your confusion.

The left-most BR is the playing bankroll that you wager from before you make your wager. The second one is the BR after the wager is made and monies returned.

The "reserve" is a side fund. What drives HMI is the money movement back and forth between the playing bankroll and the reserve.

Dave

kingfin66

06-12-2006, 08:15 PM

Hi Dave,

Thanks for posting that spreadsheet, it helps understand wha HMI is all about.

Have you ever heard of a book called "How to Make $1,000,000 in the Stock Market Automatically," by Robert Lichello? HMI reminds me of the concepts explained in Lichello's book.

It seems to me that HMI would do very well with some volatility added to the mix. For a horseplayer, high volatility would be a lower win % and higher average win mutual. If you get a couple of free minutes would you mind commenting on this.

Thanks! Looks very, very interesting.

Red Knave

06-12-2006, 09:22 PM

Have you ever heard of a book called "How to Make $1,000,000 in the Stock Market Automatically," by Robert Lichello? HMI reminds me of the concepts explained in Lichello's book.
I read that book 25 or so years ago. As I recall it comes down to simply rebalancing portfolios between equities and fixed income when there is a net 10% change in the balance. Pretty basic stuff. HMI isn't that basic.

It seems to me that HMI would do very well with some volatility added to the mix. For a horseplayer, high volatility would be a lower win % and higher average win mutual. If you get a couple of free minutes would you mind commenting on this.Dave will correct me if I'm wrong but I believe the opposite is true. HMI excels when you have a high hit rate which usually also means a low average mutual. A positive expectation is also required.

ryesteve

06-12-2006, 09:30 PM

you cannot increase your % profit with any money management system
I don't think the primary goal is % profit... it's bankroll growth...

Dave Schwartz

06-12-2006, 09:32 PM

Have you ever heard of a book called "How to Make $1,000,000 in the Stock Market Automatically," by Robert Lichello? HMI reminds me of the concepts explained in Lichello's book.

As well it should. The ideas are straight from that book, only adapted for wagering.

It seems to me that HMI would do very well with some volatility added to the mix. For a horseplayer, high volatility would be a lower win % and higher average win mutual. If you get a couple of free minutes would you mind commenting on this.

Dave will correct me if I'm wrong but I believe the opposite is true. HMI excels when you have a high hit rate which usually also means a low average mutual. A positive expectation is also required.

Actually, you are both right.

A high hit rate is important as it makes the wheels turn. However, streakiness churns the most profit.

I recall that, when using it back in the '80s, things were going so well that I would actually look forward to a few losers so the bets could get pumped up a little. I just cannot tell you the level of confidence this strategy lends to the winning player.

In 1990 and 1991 I played sparingly, and only played if I could take an entire week off (Wed-Sun) to race. During those two years I have 38 consecutive winning weeks, playing northern and southern California.

Dave

kingfin66

06-12-2006, 11:52 PM

Cool. Thanks Dave. That book has been pretty controversial for a $6.99 paperback. Check out the reviews at Amazon sometime. I have never tried it's methods with the market. I may find an equity with a high beta and see what it can do over a period of time. There are actually A.I.M user groups out there in cyberspace.

I put my recent results through the worksheet. Please understand that my ROI from 6/6 - 6/12 is very warped due to a couple of huge winners and an overall large cluster of wins, but HMI altered increased my bankroll by a huge amount compared to flat betting. Obviously my sample is very small, but the difference in $$$ won was huge. One change I made was to use 5% of BR rather than 10% as I tend to have runouts due to my style of play.

Very cool indeed.

Dave Schwartz

06-13-2006, 12:05 AM

King,

That is a typical story once they understand how to use it.

Glad to see you are having such success.

Dave

Dave Schwartz

06-13-2006, 12:48 AM

This thread has generated quite a bit of renewed interest in HorseMarket Investing. Many thanks to PA for permitting such a commercial hijacking of this thread. <G>

For those of you who may be considering ordering in the near future, please be aware that our office will be closed from June 14 to June 26 while we are away on vacation. Orders received tomorrow, June 13, will ship tomorrow. Anything coming in after that will have to wait until we return.

Sorry for any inconvenience this may cause. (Well, maybe not too sorry... I have never been on a cruise before.)

Regards to all,
Dave Schwartz

formula_2002

06-13-2006, 07:07 AM

I don't think the primary goal is % profit... it's bankroll growth...

cj wrote
I can tell you one way to improve that series, and it is something that could help almost any win better. Stop betting the chalky ones! Stick to 3.5 and up, or even higher, it will almost always improve your results.
Rye, CJ makes a good point and it must be considered in any optimal betting formula, which, I think "Kelly" does and I don’t know if Dave's does.

You must know your edge in the incremental odds range. In the example I posted, and the one Dave answered, the bettor understood that each bet had a potential to be profitable.
Questions are,:
How profitable?

Are the 1-1 shots as profitable as the 19-1 shot?

Did the bettor do a complete statistical analysis to determine his risk. ?

For optimal growth and minimal ruin, bet size has to include the functions of edge and win%.
But in the end, roi=(output)/input, a/e

ryesteve

06-13-2006, 09:40 AM

i'm wondering... how would HMI, or any other strategy, operating in a situation where bets must be placed "in bulk"... in other words, you don't necessarily know your bankroll prior to a given bet, but you do know what it is before a session of bets? One example would be betting matchups. Many players would place many of a day's bets the night before, exploiting mistakes in the early lines, and then perhaps follow up with some more bets the next day, once the players create some more inefficiencies. So, once the first bet is placed, you don't know what your bankroll will be as of the 2nd, 3rd, 4th, etc. bet, nor do you know how many bets there will be in this session (because you don't know how many line ineffiiciencies will be created). Thoughts?

Dave Schwartz

06-13-2006, 09:49 AM

Formula,

No disrespect meant here, but you are way over-complicating the issue.

You don't need to know any of that stuff. What you need is a positive expectancy.

Trying to prove that your approach works with a large sample of races is an effort in futility because your sample size will never be enough to make such a judgement.

Regards,
Dave Schwartz

Jeff P

06-13-2006, 01:02 PM

posted by Dave Schwartz - Trying to prove that your approach works with a large sample of races is an effort in futility because your sample size will never be enough to make such a judgement. Dave, no disrespect meant but I have to disagree with the above statement.

I break my own samples up into calendar year quarters. When I see an edge associated with a given approach repeated in sample after sample - quarter after quarter - year after year - including samples going forward (aka live play)...

At some point, after looking at enough samples all showing the same thing, there is one and only one conclusion to be made:

The approach in question not only works, but has a proven long term edge.

-jp

.

Jeff P

06-13-2006, 01:24 PM

HMI - or any strategy for that matter - can easily be applied to matchups...

What happens if you reframe your definition of a bet? What if, instead of thinking of each individual play as a bet, you think instead of each day's activity a bet and go from there?

Things start to get really simple. You begin each new day with a known bankroll. You then have a very good idea in advance how much of that bankroll can be risked on the next play: Today's activity. Then, by simply looking ahead into the day and counting the number of individual matchups you are about to get involved with you can decide how best to apportion the next bet (today's activity) among them. Realize going in that sometimes lines can and will move and create opportunities for you that cause you to exceed your original estimate of the number of matchups you had planned on playing. How to handle this? Easy. If you keep good betting records and look at them from time to time you'll have a pretty good estimate of the average number added plays you are likely to find on any given day. Allow for this when you make your initial estimate of the number of matchups you are going to get involved with. Then, at the end of the day, tally up wins and losses, recalculate your bankroll, and you'll know the amount of capital to be put at risk tomorrow.

-jp

.

Dave Schwartz

06-13-2006, 01:55 PM

I break my own samples up into calendar year quarters. When I see an edge associated with a given approach repeated in sample after sample - quarter after quarter - year after year - including samples going forward (aka live play)...

At some point, after looking at enough samples all showing the same thing, there is one and only one conclusion to be made:

The approach in question not only works, but has a proven long term edge.

I do not disagree with what you have posted. However, you are in a different place... You are in maintenance mode - essentially fine-tuning what you are already winning with. (We do the same thing.)

Perhaps I should have been clearer. My statement applies to those in development mode. Again and again, I see people who will not pull the trigger on an approach because they have not yet run through the thousands of races necessary to prove their approach works.

And, worse yet, most statistically-minded individuals will not play an approach built from a statistically small sample. The try to build a system by looking at "races like this one" and find that they must loosen their definition of "likeness" to the point where there is very little similarity between the races in order to get sample size because sample size is everything to them.

Dave

ryesteve

06-13-2006, 02:26 PM

You then have a very good idea in advance how much of that bankroll can be risked on the next play: Today's activity.
This is one part I'm a little fuzzy on. Let's say since matchups are a high percentage bet, you'd be looking to bet 10% of your bankroll on each race. Now, if you're betting 10 matchups, spreading your entire bankroll (10 x 10%) across those 10 races seems way too dangerous. On the other hand, if you look at the entire day as "one bet", spreading only 10% of your bankroll across 10 different bets seems way too conservative. I'm not sure how to arrive at the middle ground.

Dave Schwartz

06-13-2006, 03:11 PM

Jeff,

What happens if you reframe your definition of a bet? What if, instead of thinking of each individual play as a bet, you think instead of each day's activity a bet and go from there?

One could certainly do that. Personally, my approach to the game has (histroically) been similar to that in the sense that I make a small series of flat bets, treating the summary of those bets as a "bet."

It worked great that way.

Dave

Jeff P

06-13-2006, 04:10 PM

posted by ryesteve - This is one part I'm a little fuzzy on. Let's say since matchups are a high percentage bet, you'd be looking to bet 10% of your bankroll on each race. Now, if you're betting 10 matchups, spreading your entire bankroll (10 x 10%) across those 10 races seems way too dangerous. On the other hand, if you look at the entire day as "one bet", spreading only 10% of your bankroll across 10 different bets seems way too conservative. I'm not sure how to arrive at the middle ground.
You have to ask yourself this:

What percentage of total bankroll am I willing to put at risk in any one day? Or session?

I want to stress that every type of play, even VERY HIGH PERCENTAGE plays, carries some element of risk.

For example, I have a spot play (UDM) where strict adherence to the rules of the model results in horses selected that win their races outright approximately two thirds of the time. I don't get many plays with this - maybe four to six per week - and they seldom pay more than $4.00 to win. But I'll play them all day long given the chance.

One day earlier this year, I think it was the Friday before the Super Bowl, I was delighted to find three such horses selected as part of Piinacles's daily matchups. Based on my own previous experience with these types of plays, my probability estimate for winning each matchup was very close to 85 percent. Licking my chops, I played all three.

Can you guess what happened?

All three lost. In the first matchup, another horse ducked in front of my horse going into the far turn and he clipped heels. Fortunately for horse and rider there was no fall involved, just a lost bet for yours truly. The other two never ran a step and finsihed up the track. Further research revealed strong negatives associated with the other two that weren't part of - and still aren't part of - my model. One turned out to have been a vetscratch the week before. The other was claimed out of its last race from Scott Lake. BTW I think the guy at Pinnacle who creates the matchups used that going in. But that's a whole 'nother story.

Going by the model's inherent probabilities, the probability of losing ALL THREE could be estimated at about 1 in 295. p = 1 / (.15 x .15 x.15) - 1

Yet it happened. BTW, I'm pretty confident in my probability estimates with this type of play in a head to head matchup. And given the same situation I think I'd pull the trigger all over again without hesitation.

My point is that even very high percentage plays can and do lose. When trying to grow a playing bankroll the last thing you want to do is put too high a percentage of it at risk. Having said that I'll also say that each and every player is unique. We all have our own psychological quirks and tendencies.

For me, 10-12 percent of total bankroll is the most I'm willing to put at risk during any one playing session. Even after a disastrous string of losses I'm still 90 percent intact. I find this works out rather well for my own temperament.

The right percentage for me is not the right percentage for you and vice versa. Realizing that whatever you put at risk today has the ability to be gone tomorrow, ask yourself the following question:

How much of my bankroll am I willing to put at risk today? Then go from there.

-jp

.

Jeff P

06-13-2006, 04:27 PM

posted by Dave Schwartz - ...Personally, my approach to the game has (histroically) been similar to that in the sense that I make a small series of flat bets, treating the summary of those bets as a "bet."

It worked great that way.

Exactly.

-jp

.

ryesteve

06-13-2006, 05:03 PM

All three lost. Further research revealed strong negatives associated with the other two that weren't part of - and still aren't part of - my model.
This sounds like exactly the same path I've been down with the matchups. Using a model that had been churning away pretty nicely, I had one day where I lost 7 out of 8, and then another day a week or so later, where I lost 4 of 6. Each time was a learning experience which suggested other criteria I should be considering in conjunction with the model. Since then, all has been great... but I can't help but wonder if one day I'll hit another bad skid which will lead to even more learning.

But yeah, I see your point that it comes down to subjectivity and acceptable risk-taking. But I still have a problem with treating the day as "one bet": for example, on friday I had 4 bets, but on saturday I only had one. If I was to spread the "daily risk" over all my bets each day, I'd have wound up betting 4 times as much on the saturday horse as I did on each of the friday horses. Even though the saturday horse won, that strategy feels wrong to me. Maybe you mean something like, "I'm willing to risk 20% of my bankroll each day, and on average I have 4 bets per day, so on each bet I'll wager 5% of whatever the day's starting bankroll was"

Jeff P

06-13-2006, 05:27 PM

posted by ryesteve - ...Maybe you mean something like, "I'm willing to risk 20% of my bankroll each day, and on average I have 4 bets per day, so on each bet I'll wager 5% of whatever the day's starting bankroll was"
That's pretty close to what I meant. It's funny. For some reason I have very little problem putting 10-12 percent of my total bankroll at risk on any given day so long as my action is spread across multiple plays - matchup or otherwise. But give me the day where I have only one foreseeable play going in and I have trouble pulling the trigger. There's no way I can risk 10-12 pct of my bank on just one play. Yet I know a few players whose goals might be max bankroll growth and they wouldn't even think twice about it. But I'll usually split the difference and wind up with 6-8 percent of bankroll riding on just the one play. For some reason that feels right enough for me to be able to pull the trigger. I don't feel like I risked too much if the play loses. And I don't feel like I left anything on the table when the play wins.

-jp

.

formula_2002

06-13-2006, 05:52 PM

Formula,

Trying to prove that your approach works with a large sample of races is an effort in futility because your sample size will never be enough to make such a judgement.

Regards,
Dave Schwartz

That sounds like what I said a few years ago and it seemed most everyone, including you could not buy into it!!
But just because it can not be done doesn't eliminate it's requirement..

Jeff P

06-13-2006, 05:57 PM

posted by rysesteve - This sounds like exactly the same path I've been down with the matchups. Using a model that had been churning away pretty nicely, I had one day where I lost 7 out of 8, and then another day a week or so later, where I lost 4 of 6. Each time was a learning experience which suggested other criteria I should be considering in conjunction with the model. Since then, all has been great... but I can't help but wonder if one day I'll hit another bad skid which will lead to even more learning. I've had that same exact experience over and over over. My own strategy with matchups is highly selective. I'll generally like one side or the other AND the money line on that side enough to play 7-10 matchups each week. During the Fairplex meet the summer before last I hit a streak where I won 18 straight matchups. I knew what was coming but I just couldn't help myself. I began telling all of my friends that the linesmakers at Pinnacle were there for only one reason: Hollywood wasn't making Three Stooges movies any more! Anyone with a little playing experience should already know what happened next. Looking back at the ensuing losing streak (I think they got me for something like 7 out of the next 10) I remember asking myself how I ever managed to win money at this game in the first place. But that's the reality of putting bankroll at risk. For many reasons, it's a streaky endeavor. But when I looked at the bigger picture I realized that I had won 21 out of my last 28 plays and was a considerable amount ahead. That whole experience reinforced a valuable lesson. IMHO, sound money management is KEY for growing a bankroll while enduring those streaks - both good and bad.

-jp

.

Dave Schwartz

06-13-2006, 06:44 PM

But just because it can not be done doesn't eliminate it's requirement..

A recipe for frustration.

Dave

ryesteve

06-13-2006, 06:45 PM

For some reason I have very little problem putting 10-12 percent of my total bankroll at risk on any given day so long as my action is spread across multiple plays - matchup or otherwise. But give me the day where I have only one foreseeable play going in and I have trouble pulling the trigger. There's no way I can risk 10-12 pct of my bank on just one play.
That makes perfect sense. The chances of you losing one race are far greater than your chances of losing three, so you really don't feel like you're risking the 10-12%. You assume at least one will hit, so the "perceived" risk feels more like around 1/3 of what your total exposure really is. This is another instance to apply my "bucket of coins" analogy. If you have a $100 bankroll, and you make a $1 bet on each of the 100 coins in the bucket, yes, at the moment the bucket is tossed in the air, 100% of your bankroll is at risk, but the chances of actually losing all 100 bets is nil, so the "risk" in terms of what you conceivably stand to lose, is far less than your 100% outlay.

I began telling all of my friends that the linesmakers at Pinnacle were there for only one reason: Hollywood wasn't making Three Stooges movies any more! Anyone with a little playing experience should already know what happened next.
Hahaha... exactly. After the streak I've been on (nothing compared to your 18 in a row, but decent) I know I should be escalating the bet size a bit, but I can't help wondering if I'm due for a "correction". I haven't been at the matchups long enough to get any kind of a feel for the ebb and flow.

formula_2002

06-13-2006, 06:46 PM

A recipe for frustration.

Dave
Or,perhaps..adventure..have a nice vacation :)

Vegas711

06-14-2006, 06:45 PM

The first thing you need to do is record a couple of hundred bets, from this data you will be able to formulate the best wagering sequence that will maximize your bankroll growth. Roi means very little it is the rate of growth that is inportant.

If you find that 80 % of the time following a win you lose a race then you will know that it is best to regress your bet size in the race following a win. If you seldom lose more than 4 races you can then do a increase of x dollars following Y loses.

Daves work looks good from what I have seen on his web, I would recommend a little experimentation with adjusting bet size with the odds you get.Maybe he covers this? from his example which shows betting only 2-1 price horses if I were getting 6-1 instead of the 2-1 I would look at betting a couple of dollars less than what may be listed in the charts becouse of the preminum in odds.

I agree also with Dave that Flat betting is not the way to go, the POWER of increasing Bankroll Growth depends upon aligning bet size with your wins, if you almost always have your small bets on lossing races and large bets on winning races you will win. Every bet I make is on a horse that I have a very strong opinion on does this mean that I win every race, NO. I know from my betting records that the most winners I have had is 6 in a row most of the time its 1 or 2. After I hit 3 in a row I begin to cut my bets way down, this has saved me a lot of money.

In summary align you bet sizes with your cycles of wins and losses, find from studying your betting records the BEST FIT.