Response to Game Theory (Beware: long post)

[tanda]

Response to theoretical discussion contained in closed "DRF on Odds Plunging" thread:

Game Theory:

I agree with many of your points, but they are somewhat non-responsive to the issue at hand. That is my fault since I did not do the best at explaining my theory.

Let us assume a 50% win rate with a 50% R.O.I. with bets on all races on a circuit by “Super Handicapper” who we will nickname SH. You and I agree that a 50% win rate on bets on all races would lead to a 50% or higher R.O.I. (since SH would be winning with numerous non-favorites because favorites do not win 50% of all races). We also will assume that this is a real and established level of ability and not due to short-term fluctuations.

Let us assume a “long-run” of 250 cards a year, 9 races a card and 20 years of wagering. Thus, 22,500 wins in 45,500 races with an R.O.I. of 1.50. Average odds on winners would be 2.00:1. Let us assume track takeout of 17% and dime breakage.

SH winners win at a 50% rate, but they have only 27.2% of the money wagered on them. Huge overlays. Thus, huge overlays are winning 50% of the races. Do you realize the problem?

Continuing on from another perspective.

With the takeout and breakage assumed, the average horse will return about 0.82. The average horse is an underlay. In fact, it is possible that all horses are underlays (but highly unlikely). But since the average horse is an underlay, it is a near certainty that a majority of horses are underlays. In a typical eight horse field, I estimate that 1-2 horses per race, on average, are an overlay. This figure is consistent with the belief of others. Let us assume that the number is 2 making 25% of the field overlays. I believe this number is slightly high, but we will use it anyhow. A lower figure strengthens my argument.

Returning to my question above: do you realize the problem, now?

SH cashes bets on overlays in 50% of all races, yet overlays constitute only 25% of the field. I am unaware of any theory that suggests that being an overlay, in of in itself, allows horse to win more than their fair share of races. Therefore, in the long run, if overlays constitute 25% of the fields, they should win 25% of the races. SH cannot be winning on overlays in 50% of all races since overlays do not win 50% of all races.

To address potential criticisms of my logic.

Note that I have assumed that the 50% winning bets by SH were all on overlays. A critic might argue that not all winning bets need be on overlays in order for the total return to be positive. So the figure may be lees than 50%.

I do not consider that a serious criticism for three reasons.

One, SH cannot afford, if he is maintaining a 1.50 R.O.I, to bet on many underlays. Even betting 20% of the time on underlays makes an overall 1.50 return much more difficult. To return 1.50, the average expectation on all wagers must be 1.50. Any wager on a horse who is an underlay, for example 0.90, must be counter-balanced by a wager on a horse with a 2.10 expectation or two at 1.80, etc. to bring the average to 1.50. It seems that SH will not be able to find these other overlays sufficient to offset more than a tiny amount of underlaid wagers.

Second, although SH may win some bets on underlays, there also may be overlays who win without SH having bet on the winner. Thus, the number of overlay winners includes both overlays wagered on by SH and those not wagered on by SH. The estimate of 50% overlaid winners discussed above may be reduced a tiny amount by those SH winners who are not overlays but will be increased a tiny amount by those overlay winners who SH does not bet. Thus, the estimate of 50% should remain fairly constant.

Third, even if we reduce the estimate by a significant amount, for example to 40%, we still have overlays winning much more than their fair share of races.

Now, I will address the problem from another perspective.

Let’s imagine that a genie arrives and informs you that 25% of all horses are overlays and they win 25% of all races. Better yet, she knows with complete accuracy who those horses are. She will provide you with this information. On average, you will, based on an average field of 8 horses, wager on 2 horses per race and bet the winner in 25% of the races. You will show a profit since all bets were made on overlays, i.e. horse with positive expectation.

Everything is great. Then, you are infected with a virulent strain of the “I gotta cash now” disease. In the past, you saw bettors winning a higher percentage of races than you, but you took solace in the fact that they were betting on too many underlays and you were profitable and they were losing money. Unfortunately, now that you are disease stricken, the disease causes you to lose focus on the profits and instead concentrate on picking additional winners.

How are you going to boost your win rate (12.5% of bets, 25% of races)? Obviously, you must start wagering on additional horses. By definition, all of those other horses are underlays. Thus, the only way to increase your win rate is to bet underlays. There ain’t enough overlays winning to allow you to increase it any other way. Remember, you were already betting all the overlays.

This means that there is a limit to how many races a handicapper can win, betting every race, without hurting returns by adding underlays to the portfolio. Based on the takeout and breakage, that number must be significantly less than 50%.

Finally, I have a third perspective that may be the simplest yet since it does not even consider the concept of R.O.I.

In the long run, what do you believe the winning percentage is of the most likely winner in a race? I mean the truly most likely winner.

The lower limit on this number must be 33% since the public’s top pick wins 33% of all races. Is this number also the high limit? I suspect not. My estimate is 33%-40% with 35%-37% the most likely. In other words, the truly most likely winner of a race, on average, should win about 35%-37% of the time. This number is consistent with most everything I have read and learned about the game. I know of no handicapper, particularly those who make oddslines, who, again on average and assuming all races are handicapped, assign more than this estimate to their top pick. Remember, we are discussing their top pick in every race run, not just selected races. I have no doubt that selective handicappers identify situations where the most likely winner wins more than the estimate above. We are discussing handicapping every race.

Assuming the above is agreeable, do you understand why it is then impossible for SH to pick the winner in 50% of all races.

SH has identified a class of horses (his most likely winners, one per race, every race) who win 50% of the time. But, we have agreed that the truly most likely winner of a race, on average and considering all races, does not win that percentage of the time. The above estimate puts an upper limit on the number of winners that can be picked if all races are played. If the truly most likely winners in all races win at a rate of 37%, how can a collection of horses (SH’s selections) who at best (even if he always correctly identifies the truly most likely winner) out-perform the 37% rate? At best with perfect handicapping, he can only match the 37% rate.

This concept has never been discussed in the past, to my knowledge. It is the idea that there is an upper limit to how many winners a person can pick if all races are played, no matter how skillful they are. If picking winners is your goal, ignoring R.O.I. considerations, then optimal performance is to correctly identify the truly most likely winner. Your performance will then match the win rate for the truly most likely winners. This is the optimal performance, it cannot be bettered. So, to believe that SH can pick 50% winners playing every race, then you most believe that every race run contains a most likely winner who average will all other most likely winners in all others races wins at a 50% or greater rate.

The ramifications of my 33%-40% estimate being wrong (an estimate I believe shared by most) is huge. If it is in fact true that the most likely winner of a race win 50% of the time, then all software developers, handicappers, etc. who make oddsline and other value-oriented betting decisions are committing gross errors. All of these people assign much less than a 50% chance to the top choice on average. They assign an amount consistent with my estimates. If our estimates are wrong, then we are severely underrating the chances of the top choice and, consequently, over-estimating the chance of the remaining field. The mistakes are enough to make any oddsline created with those mistakes fatally flawed regardless of its other strengths.

Can a handicapper win 50% of selected races? Absolutely. One example is to only bet 0.10:1 horses. They win well over 50% of the time.

Can a handicapper return 1.50 R.O.I. on selected races? I believe so, since I believe that some horses (more rarely than most others believe) do have winning chances and off odds sufficient to create a 1.50 expectation.

Can a handicapper do the first or both of these things betting every race? I believe not. Not because of a lack of skill, but because the structure, nature and mathematics of the game set result limits even with optimal play and the results at issue are above those limits. Even with perfect play, there are limits to results in all fields. Handicapping is no different.

[Rick]

tanda,

I have to agree with you. Published records of the most highly regarded handicappers playing all races are rarely able to achieve more than 30% winners and even then, usually at a loss. But it makes sense to believe that some are considerably better than that and, I know for a fact because I've done it myself, that you can win better than 30% and still show a profit playing all races. But, for me, it always winds up being more profitable if I only play about 1/3 of the races because that's where the profit is. I can imagine using alternate methods for selecting overlays in the other 2/3 of the races but I can't see winning 50% of them either after I've already eliminated horses with 30% chance as underlays. But we're talking about overall averages here. It is possible that there is some track out there (maybe it's Calder) that has an unusually high percentage of winning favorites that is so easy to handicap that one can get 50% winners and still show a profit. If you have 4 horse fields, 50% is the average win % for favorites, so it may be possible somewhere. But I'm not a handicapping guru and I'm not writing books or giving seminars, so I might be wrong.

[JimG]

Rick:

Sorry to be off topic here, but like a dunderhead, I just realized that you used to post as "Rick Ransom". I wondered what happened to him and glad to see you are still here.


Jim

[boxcar]

Re: Response to Game Theory (Beware: long post)

Yes, a long post, but one the finest I've read in a long time on the game -- very well thought out and carefully constructed for such a closely worded argument.

Boxcar

[baravot]

tanda,

A well-reasoned and strong post. I appreciate your reasoning skills.

[Rick]

Jim,

You're correct. After not posting for a while due to being overwhelmed by family business, I was denied access when I tried to get back in for some reason. So I just signed up again. I suppose I should demand credit for all of my old posts, but by now I think there are others who have far surpassed my numbers.

[GameTheory]

Tanda,

Firstly, I agree 100% with your post (the reasoning & the conclusion), GIVEN your basic assumptions about the possible "upper limit" you were talking about.

The upper limit stuff is key, and that is EXACTLY the same thing I was getting at in my original post when I made up that example about the 3 horse race, and the true chances of the winning horse in each case being close to 100% even though the stats would seem to show each one having only a 33% chance.

To quote my conclusions:


There are a couple of basic scenarios:

-- the public vastly over-rates & under-rates certain horses in each race ALL THE TIME

-- the public handicaps quite accurately in every race, and big errors are rare

The thing is, both scenarios will give you the EXACT SAME SET OF STATS, because the first one averages out, and the second one doesn't need to.



So, you could be correct, and that the true "upper limit" is around 37% (the second scenario), but there is actually no way to tell, is there? We can be sure of the lower limit, but the ONLY way to find the upper-limit is to find the highest percentage handicapper in the world betting every race, right?

Your numbers sound reasonable, but it is quite possible they are dead wrong just from a theoritical basis, right? The whole argument rests on the assumption that you know around where this upper limit is at, but that assumption is really just a reasonable looking guess, and based on averages. (Certain parts of your argument rest on the fact that lots of other people agree with you.)

I am glad you mentioned that basically no one has ever talked about this before, despite claims that we are beating a dead horse.

So, my question to you is, what happens if you were to be confronted with a real-life Super Handicapper, and verified his performance to your satisfaction? Having met such people myself (more than one, and at different tracks) and seen them perform, I simply cannot accept your conclusions, which are perfectly reasoned but apparently based on false assumptions. I cannot believe otherwise, because your conclusions cannot be reconciled with the real-world evidence I have. (Although these people would skip 1-2 races a card -- races with lots of FTS, etc. -- so not quite so idealized, but our resident potential SH said he does the same.)

But putting that aside, and just dealing with the theory, let's go back to the two scenarios, putting them another way:

Scenario 1: the upper limit of what is possible is quite high, much higher than most people think.

Scenario 2: the upper limit is just a few points higher than average (just a bit more than the amount of winning favorites)

If Scenario 2 is true, then you are 100% correct.


So, the heart of the matter:

Is Scenario 2 really any more likely to be the truth than Scenario 1? Why?

Remember, they LOOK exactly the same in the aggregate.

Isn't it possible you are just making a conservative guess, assuming the deviation from the average isn't that great? Isn't it possible that this is the normal impulse of most people?

Considering the fact that (as you mentioned) no one has ever discussed this in public or in print before, that these unthought-out assumptions are just plain wrong? A case where intuition has failed us? Like the Earth being flat, no?

Given that, we might consider shifting focus away from the theory for a while to the process of collecting evidence. Try to find that real-life Super Handicapper before we conclude he doesn't/can't exist.

Leaving Karl out of it, am I the only one who has ever sat at the track day after day with a guy cashing an ungodly amount of tickets race after race? Any other witnesses out there?

[Dick Schmidt]

Game Guy,

I have done it; sat next to a super handicapper that is. Not just for a few days but for about 3 or 4 years, 2 or 3 times a week. His name is Tom Brohamer, and he could just flat pick horses. I don't have access to his betting records, and my memory can't supply exact numbers, but I would estimate that he hit about 60% of his bets, playing one horse almost always, and playing about 7 races a day on a 9 race card. He didn't like the Pick-6 much and only played it about once a month, never for more than $100. I personally saw him take down three that totaled well over $100,000. One for $79,000 was at Pomona (Fairplex) back when it was a hand timed track that "everyone" said was unbeatable because the times were so far off.

If I could do what he does, I'd toss out all my computers and die happy.

Dick

[Triple Trio]

Originally posted by Dick Schmidt
... His name is Tom Brohamer, and he could just flat pick horses. I don't have access to his betting records, and my memory can't supply exact numbers, but I would estimate that he hit about 60% of his bets, playing one horse almost always, and playing about 7 races a day on a 9 race card.

From p.43 of Dick Mitchell's book "Common Sense Betting":

"I have the pleasure of knowing a good number of winning handicappers, and not one of them can pick more winners than the public. That is, I don't know a single handicapper who can bet every race for a long period of time and achieve a 32 percent win rate. In fact, most can't achieve this win rate even by being selective. Tom Brohamer and James Quinn can boast win rates over 32 percent. The reason they're able to achieve this is because they're highly selective about the races in which they choose to invest. Normally they get between one and three plays per day at a single track, averaging about two plays per day. Mark Cramer has a 28 percent hit race, as does this author. The difference is average mutuel. Mark and I average more than $12 while Quinn and Brohamer average $8. The key to this discussion is "edge"."

[cj]

It would seem to me that if a handicapper has come up with a method which consistently hits 50% winners from all races, his definitions of overlay and underlay would be quite different than those proposed in the original post here. If someone is able to pick this many winners, the prices would be irrelevant, as surely he would be a winner over any reasonable length of time. I'm sure there is an occasional race card where selecting half the winners would produce a loss, but in the long run, you would always come out ahead.

I honestly don't see why acheiving this rate is so hard for some to believe. My method of play would never give me such a high percentage, but picking 1 to 2 more winners than average per day seems within the realm of possibility to me.

CJ

[Rick]

And there are those who would swear that Sartin picked 100% winners. Sometimes it's hard to separate what one believes from what's really true. If that weren't true, magicians would all be out of work.

[cj]

Originally posted by Rick
And there are those who would swear that Sartin picked 100% winners. Sometimes it's hard to separate what one believes from what's really true. If that weren't true, magicians would all be out of work.

I don't get it. Are you saying you don't think it is possible? Just curious.

CJ

[tanda]

Random thought:

1) The best documented performance would not establish an upper limit, but just that the upper limit is at least that high,

2) I am not going to dispute Brohamer's results per se, but I have never read him, Quinn, Mitchell, etc. claim that he won 60% betting one horse and playing nearly all races. Plus Dick's post acknowledged that he did not play every race and sometimes played more than one horse. So even if the results are accurate, they differ from the assumptions I made: bet one horse in every race carded. I do believe (at least on dirt races with little or no FTS) that Brohamer would be one of the likeliest candidates to come closest to the optimal win percentage (assuming that was his goal, which is different from making a high R.O.I.)

3) My point is that the upper limit, if all races are bet, is the percentage of times that the most likely winner wins. Optimal performance, in picking winners, not R.O.I. necessarily, is to bet the most likely winner, thus, optimal performance can never exceed the win rate of the most likely winner. Most handicappers seem to think that the average most likely winner, considering all races, wins between 33% and 40%,

4) As to whether a 50% handicapper, betting all races, would change my theory, the answer is no, the theory is just that the upper limit is defined by the most likely winner win percentage. A 50% handicapper would merely change my estimate, probably shared by most handicappers whether they realize it or not. That is why I wrote that those results would alter some widespread conceptions of the relative chances of horses in a race. As I wrote, most value linesmakers, including software, does not give the most likely winner a 50% chance on average. Most average 33%-40%. If 50% if the correct average, then almost all oddslines are off for all horses by a significant degree ... and maybe fatally flawed regardless of their other strengths (presuming the betting line is used to make the betting decisions).

[cj]

Originally posted by tanda
Random thought:
3) My point is that the upper limit, if all races are bet, is the percentage of times that the most likely winner wins. Optimal performance, in picking winners, not R.O.I. necessarily, is to bet the most likely winner, thus, optimal performance can never exceed the win rate of the most likely winner. Most handicappers seem to think that the average most likely winner, considering all races, wins between 33% and 40%,


This is where I have a difficult time with the argument. It can never be proved, or even tested, as the "most likely winner" can never be identified before the race, or even afterwards.

CJ

[Triple Trio]

Originally posted by cjmilkowski
It would seem to me that if a handicapper has come up with a method which consistently hits 50% winners from all races...I honestly don't see why acheiving this rate is so hard for some to believe.

For those of us who treat horse racing as probabilistic events, it's quite hard to believe. Unlike Tiger Woods, Mike Tyson, Kasparov, etc., we handicappers don't have that much control over our own destiny. We can guess but will never be sure how a jockey is going to run a race; the horse we pick may encounter trip problems; the pace scenario may turn out to be completely different from what we expected due to some unforeseen events; the photo finish may not go our way... So even if we've picked the right horse, it doesn't necessarily follow our horse will win the race. There are so many random things that can cost us the race. Plus there are races with shippers, first time starters, etc. that are full of unknowns.

What's more, we are not talking about achieving 50% win rate over a short period of time, but consistently over a long period of time. This is equivalent to Tiger Woods winning the 4 majors not just in one year, but year after year for many years to come.

[cj]

Originally posted by Triple Trio


...the horse we pick may encounter trip problems; the pace scenario may turn out to be completely different from what we expected due to some unforeseen events; the photo finish may not go our way... So even if we've picked the right horse, it doesn't necessarily follow our horse will win the race. There are so many random things that can cost us the race...

All of those things should even out over time and give us winners when we pick the "wrong" horse just as often.

Keep in mind, this all goes back to an old thread and is talking the Florida circuit, with very few shippers, and the original "SH" said he passes races with 2 or more 1st timers.

CJ

[Triple Trio]

Hello CJ,

You asked "why acheiving this rate is so hard for some to believe". The reason is because (1) we haven't seen any handicapper who has done this, and (2) from a theoretical point of view it is difficult to achieve.

Now it may be possible that someone has actually done so without our knowing. OTOH, it's a basic tenet of scientific reasoning that the more extraordinary a claim, the stronger a proof is required. Absent such a strong proof, I prefer to remain sceptical. It doesn't necessarily mean that I believe someone is lying because I don't see the motive for doing so. It's just that I am in a state of puzzlement.

TT

[cj]

TT,

I'm just basically playing devil's advocate. I have no idea how likely it is to acheive this feat. It would seem plausibe at a few tracks, particularly those that are plagued by short fields.

CJ

[rrbauer]

CJ wrote:

as the "most likely winner" can never be identified before the race, or even afterwards

If you're saying that ("never") as a generality, then I disagree with you.

There are many races where the "most likely winner" can be identified; where one horse just lays over the rest of the field; where the pace scenario is obvious and so is the beneficiary of that scenario. If you're saying that there are times where the myriad of handicapping factors and associated nuances are such that the "most likely winner" becomes a subjective proposition then I would agree that identification of such is questionable.

I believe that Tanda has offered up this classification in a conceptual sense to establish a theoretical outer limit; or, gauge, against which to measure the liklihood of achieving certain levels of success. And, he's left it with a generous tolerance range so that we can plug-in what we believe to be reality.

There are exceptions to most rules. When I say that there is nobody betting into the pari-mutuel pools at the major racetracks in the U.S. today that is betting every race on the card using a single selection AND hitting at least 50% winners, you may agree or disagree. Opinion for opinion. Tit for Tat (but none for Nat...old joke). If that assertion can be disproved with specific evidence then I will have a new hero(s)!

[Rick]

CJ,

No, I'm not saying it's not possible but we've all at one time or another believed that something is better than it really is. Sometimes it's because of fraud, sometimes because of small samples, sometimes because of our misinterpretation of the results. Dick believes that Brohamer can pick 60% winners and I believe that he believes that. Others believe he can pick no more than 32% winners. They can't both be right even though they both have an equal belief in their position.

Some believed that Sartin could pick winners in every race but it was really a trick. That was an extreme example, but one could easily believe that someone had 60% winners on one horse if you only saw the one winning ticket out of the two that he selected. Just an example that doesn't require anyone "lying" about anything.

The only important thing to me is that people don't get the idea that these kinds of levels are what they should expect themselves. I think many good horseplayers have been ruined by trying to achieve unrealistic (for them) levels of winners. If I can't do it, I've never seen anyone else do it, and you can't teach me how to do it, then I don't care what you say you or others can do. It's just totally irrelevant and I don't understand why people would continue to argue about it.

[Triple Trio]

Originally posted by cjmilkowski
It would seem plausibe at a few tracks, particularly those that are plagued by short fields.

Your comment reminds me something I read a few days ago. One of the Chinese papers here ran a special on horse racing in South Korea. One of the things mentioned is that the favorites there win about 50% of the races. I have no idea how true this is. Let's assume it's true. So it may just be the case that a very set of circumstances may make it easier to pick winners at certain tracks. Even so, if the pools are pari mutuel, I don't think you can make money picking 50% winners.

[Rick]

Triple Trio,

Maybe that's why I could never win at the track when I was in Seoul. Heck, it took me long enough just to figure out which numbers meant what in their version of the racing form. I will say that they have huge crowds there though.


To continue with the subject of whether 50% (or is is 60% or 80%now?) is attainable, I think that our opinions are usually influenced by whatever information we've used. While it seems unlikely that one could get anywhere near these levels using only information available in the past performances, it's entirely possible that the situation could be improved greatly by adding other information. My candidates would be visual handicapping (body language), trainer stats indicating intent, toteboard information, post position/pace advantages at various distances, and occasionally equipment differences. You could also add private workout info if that's available on your circuit. So, there are a number of ways to increase your win % from the usual 30-32% level but you must work with all of the available information and know your track really well. In my opinion, based on regression studies I've done and seen, getting over about the 40% level would probably be impossible without using some subjective information, the most valuable of which is probably the body language factor.

[Triple Trio]

Originally posted by Rick
it's entirely possible that the situation could be improved greatly by adding other information. My candidates would be visual handicapping (body language), trainer stats indicating intent, toteboard information, post position/pace advantages at various distances, and occasionally equipment differences. You could also add private workout info if that's available on your circuit.

In a way, this has already done here in Hong Kong. The big computer teams' software contain over 100 variables extracting all available info from the past performances. They also have guys watch video replays to adjust the beaten distances. There are also subjective specialists who deal with morning workouts, paddock parades, etc. They also incorporate toteboard info into their final odds. Even so, they still have to suffer losses from time to time. If these teams were to move to North America, I don't think they can achieve 50% win rate playing all races carded.

[GameTheory]

The Hong Kong teams are not selection oriented, of course. They calcuate fair odds, and bet overlays in all pools. And the pools are enormous.

They are also only dealing with something like 700 races a year and 1200 total horses, right?


A year's worth of data for North America covering all tracks and you're talking about something like 60,000 races & 70,000 total horses. With relatively tiny pools.

Much different ballgame.

[Rick]

TT,

Are there any published accounts of what the 100 variables are that they consider significant? I have seen a paper on a 20 variable model based on their data that was interesting. Some variables that were significant in that model, such as number of career starts, don't appear to be too important in US racing.

[Triple Trio]

Originally posted by Rick
TT,

Are there any published accounts of what the 100 variables are that they consider significant? I have seen a paper on a 20 variable model based on their data that was interesting. Some variables that were significant in that model, such as number of career starts, don't appear to be too important in US racing.

Not that I am aware of.

My understanding is that number of career starts, jockey change and weight change are some of the most, if not the most, important variable. Personally I wonder why so many variables are needed. If the major factors account for say 30-40% of the results, then some of the remaining factors must account for less than 1% of the results.

[karlskorner]

I go away for 4 days, come back and it's still going on under a different post. To add to your post above, I also do not play a switch from grass to dirt along with 2 or more FTS. there are 1 or 2 other type races I don't play.

For the past 2 hours I have been reading what I missed and in my mind it's all boiling down to one thing. 90%of the posters are playing from home or OTB, a small group attend the track on a regular/daily basis. Yet I keep reading over and over that it is impossible to do what I say can be done by any member of this board. Dick knows a few that have and are doing it, some others know of 1 or 2, but the majority only know from what they have read or from their own experience. Do the proven handicappers at any track/OTB wear a sign that says "Hi, I'm a handicapper, ask me how I'm doing". Aside from the nut who runs around screaming "I had the Tri", you won't be able to identify a handicapper unless you go up and ask 'are you a winning handicapper?", be sure to write down the answers. Unless you have personal knowledge and have actually spoken to at least 100 winning handicappers, it means nothing to me to state that mathematically this or that can't be done, or my data base stats says it has to happen this way, it all proves nothing. Go get the answers from the "horses mouth" If somebody keeps telling you over and over that something won't work and they have the math or stats to prove it, after awhile you will start to believe it.

Karl

[Derek2U]

This is my question for all you handicapping kingpins:

Why not use more than 1 system based upon the odds?

Let's say that most winners, paying $8.00 Max and Less probably
share some past-performance traits in common. Let's also
assume that maybe Winners, paying between $8.01 and $12
form another sub-set. Etc Etc ....

Why dick around with Value & Morning Odds Lines when you
can apply various prediction systems that were based on
the winners payoff & see which group each horse is selected by?

If your longshot system pick Horse "A" and you look up and see
he's 3:1 then dump him. Etc Etc ....

In this way, you can "scientifically" obtain your prediction systems
based on the odds over many, many races and not try to use 1
system, realizing that the "obvious" horse is unlike a longershot
one. Comments please.

[Rick]

Derek,

That's pretty much what I said before. There's nothing preventing you from using more than one technique to find overlays. If one doesn't work, use another. But most of us use only one method and that's why there's a limit on what you can do. Correct me if I'm wrong, but I'll bet Karl uses a variety of techniques to find winners at the one track that he specializes at. Others use the much easier approach of finding a small percentage of easy-to-beat races selected from many tracks.

Really though, a lot of this is like religion. I learned a long time ago not to argue with people about things they really believe in. I can't prove that the miracles mentioned in the Bible didn't happen even though I don't see them today. They might really have happened and there might be a reason why we don't see them anymore. One could say that someone who's seen one of the really good magicians today would probably believe they performed miracles if someone didn't tell them differently. How about UFOs. Are there people visiting us from other planets? Can't prove it one way or the other. Psychics, ghosts, etc., can't prove or disprove it. Most people don't believe in the guru handicapper either but that's no proof that they don't exist. All we have is what is usually called "anecdotal" evidence which really means that somebody really believes it's true. Telling a guy with a reputation that he has to prove to you that what he claims is true is like telling Jesus he'd have to show you one of those miracles before you'd buy into his religion. A lot of guys swear he really was that good, so you'll just have to either believe or not.

[Derek2U]

I really think that any1 who uses some arithmetic (mathematical)
prediction systems should re-think his position. You will NevER
PICK longshots unless your system is geared to detect them.
heheH ... I know I've bad-lipped "Value" over & over but I got
good reason: Value is Intimately fettered to one's system itself:
and so, it's not geared to see the longshots. Horseplayers who
thrive on math & puters & databases should follow ploys, techniques that us finance guys use --- not because we finance
guys have brains but because we just use techniques that firms have spent millions 2 perfect --- in other words like the firm i work for has more Ph'Ds in physics/math/puters than most USA colleges have. These Drs R Big Brains making even BiggER $$$ ... hehe they use so many things that handicappers w/ data bases never even heard of .. like u all think those HongKong guys are
great .. Bull S ....

[rrbauer]

Karl said
Unless you have personal knowledge and have actually spoken to at least 100 winning handicappers, it means nothing to me to state that mathematically this or that can't be done, or my data base stats says it has to happen this way, it all proves nothing. Go get the answers from the "horses mouth"

Comment:
Tell you what you do Karl. You go talk to the 100 winning handicappers (actually 30 would probably provide a valid sample) and get it from the "horses mouth" and let us know what you find out, because you're the one who seems to need closure on the issue. You say it's being/been done. Prove it! (And leave out the third-party hearsay).

[Rick]

Derek,

Yeah, I worked with a lot of PhDs too during my 20 years doing military and physics simulation but it had very little affect on my handicapping. I doubt if the financial savants are much better. It is possible to use mathematical systems combining underbet factors to show a very good profit at reasonably high win percentages. But you can't just throw together a bunch of obvious factors and run a regression or neural network on it and get any good results. It takes many years of research to select the right factors that don't duplicate the way the public bets. After the hard part is done, it's relatively easy to calculate a rating and bet accordingly.

But you are correct that it's much easier to develop a winning system based on longshots. That was the way I won in the 80's and many of the systems still work despite changes in public betting habits.

[Dick Schmidt]

Financial market PhD's have the right stuff, huh? That must be why their track record over the years on Wall Street is so good. And since all the firms have them, that must be why the market is doing so very well for all the traders and investment banks. Yup, no doubt about it; they've got ALL the answers.

Dick

[karlskorner]

rrbauer;

I have nothing to prove, as I have RRBAUER, who declares to all that he is a "two hands bettor", which imples that he is a heavy hitter, in order to be such, he must be able to recycle monies that he has won. Closure, hardly, all I have to do is interview RRBAUER and pass on to all what I have learned from the "two hands bettor"

Karl

[smf]

Originally posted by karlskorner
. Aside from the nut who runs around screaming "I had the Tri", you won't be able to identify a handicapper unless you go up and ask 'are you a winning handicapper?", be sure to write down the answers
Karl

How true! Everyone knows that the proven handicappers are the ones running to the parking lot after the 6th race screaming "I'm leaving after the 6th race! Does anyone wonder why?! It's because I cashed win 5 tickets!".

[Rick]

I might be missing something, but I don't think that leaving early is really a smart move. If the method you're using has picked up on a bias, why not continue to take advantage of it. It sounds superstitious to me, and superstition is for losers. Just my opinion, I could be wrong.

[smf]

Rick,

Karl posted in a thread in this section ('DRF on odds plunging') that he left da building at Calder after he hit 5 of 6 winners which he claims is why he was logged on the board earlier than usual that day.

I'll allow karl to answer your comment> ...."I don't think that leaving early is really a smart move. If the method you're using has picked up on a bias, why not continue to take advantage of it. It sounds superstitious to me, and superstition is for losers. "

That statement was his (leaving early), not mine.

[cj]

Not many things in life are certain, but this one is...if I hit 5 of the first 6, I'M STAYING!

CJ

[Rick]

smf,

And your point is .... ?

[smf]

cjm,

Same here. <G>

I wouldn't leave until I didn't cash at least two tickets, LOL.

[Derek2U]

does anyone think that winners paying $8.00 or Less differ
from winners who pay $15 or more? In other words, has anyone
examined Winners & how they measure up on factors? If so,
wouldn't it then be easy to compare horses with those profiles?

[cj]

Originally posted by Derek2U
does anyone think that winners paying $8.00 or Less differ
from winners who pay $15 or more?

Derek,

Yes, I think they differ by at least $7 :p .

Only kidding of course. When horses start paying in the $15 range, I would think the factors that would point to them would be a wide variety, but worth looking into, maybe by class and distance of race.

CJ

[Dick Schmidt]

About longshot winners,

It seems to me that most double digit winners have at least one or two minor flaws, just enough to give most players an excuse to run to bet the favorite. True longshots, say $20 or $25 and up will almost always have a major flaw, glairing up from the PPs and scaring off most everyone. The trick, I think, is to ignore flaws at long prices and look for a positive or two. I find a lot of long priced winners this way. Of course, I usually fudge and include one of the three low odds horses in my bet as well. You lose a lot more than you win with high odds horses, but man, is it sweet when they come in.

Dick

[Rick]

Dick,

Are you in favor of the old William Scott idea of selecting one of the top 3 favorites?

[Dick Schmidt]

Rick,

No, not really. Top three generally win around 65%, so one of them figures to be a solid contender, but I bet what I consider to be the fastest two horses who are in good form. The lower the class of the race is, the more important form is. Sometimes I bet two longshots, sometimes two lower priced horses if I can get even money on the total bet. I just find that most of the time one of the top two or three is one of my horses.

Dick

[Rick]

Dick,

That's pretty much what I thought, just wondering if that idea had any merit now that the short-priced horses are no longer underbet to any large degree. Does your definition of form include recency? An awful lot of my overlays these days are laid off horses that the public seems to hate to bet.

[Dick Schmidt]

Rick,

In a word, NO. I never look at any sort of a Past Performance Chart, and have no idea of when a horse last raced. I know I won on a horse the other day that had been off more than 2 years. I heard the announcer mention it when the horse crossed the finish line.

Dick

[Rick]

Dick,

So, by form you're not referring to "current" form, or condition? I've tried to use that kind of thing and haven't found anything that helps more than it hurts. But I feel like I must be missing something.

[Dick Schmidt]

Rick,

What I am looking at is a comparison of how the horse ran in its last few races compared to it's overall (last 10) races. In HSH, the one number that best represents pure overall speed is a rating that Dave developed some years ago that he calls a Power Rating. The program averages the last 10 Power numbers, then compares the last five races to the average on an individual basis. If the average is 80, and the last race the horse ran an 85, then the form rating is +5. With a bit of pattern recognition (plus some help from the program with moving averages) I find that I can identify horses who are improving or holding good form.

There are two major aspects that are missing, though both are available in HSH if I would take the time to look. They are recency and class. When I'm beaten by a horse that pays $3.20 or so, I sometimes look to see what the crowd saw that I didn't. Almost always it is a big class drop. These horses never pay well and don't win their fair share or even close to it, so I don't mind at all. All the money dumped on them pushes up the prices on my horses when I do beat them. I think that recency is way overbet by the "Racing Form" handicappers. Again by ignoring it, I lose a few, but my ROI goes up.

At one time, I wanted to be a big time handicapping stud and hit everyone over the head with my 75% win percentage. I've calmed down a lot since then, and am now a very relaxed player who hits about half his bets (both figures using 2 horses per race). Before I played about half the races on offer at one track and wouldn't blink at betting two horses at 8/5 and 2/1. Now I play 7 or 8 tracks and won't take less than even money on the dutch, usually much more. By ignoring most of what everyone else uses to handicap with, I find my ROI has gone way up and I only spend at most two minutes per races picking horses.

Dick

[Derek2U]

Nothing is more important, either as a solo factor by itself or
as a contributing factor. I place it up there with year-end
bonuses; but the difficulty is this: You can't be 100% sure when
(if) a horse is loaded for bear. Yet not giving it respect will most
definitely lead to financial ruin.

[Dick Schmidt]

The message is you don't need to be right 100% of the time. I'm only right on about 25% of the horses I bet and make a comfortable living. I used to advise people at seminars that to make more money, they needed to learn to lose more.

Dick

[Rick]

Dick,

What do you think of form evaluation based on pace figures, along the lines of what's mentioned in "Blinkers Off"? It sounded logical to me, but some limited testing didn't improve my results.

[Dick Schmidt]

Rick,

Maybe we should change the name of this thread to the Rick and Dick Show.

Anyway, to answer your question, I have no idea. Never read "Blinkers Off". Personally, I wouldn't use pace figures. There are too many horses that are strong in the early going, but fade time after time. I think you need an overall rating, but not a final time rating either. Doing it the way HSH does, you compare the horse against itself. Once you decide it is in good form, you must then decide if it is fast enough to beat the bunch it's up against. A slow horse at its absolute peak is still a slow horse.

Dick

[Rick]

Dick,

We really are off-topic here, but isn't that the way all of these threads run now? Anyway, thanks for your thoughts on the subject. The reason I mentioned that book was that so many, including Ziemba (Dr. Z), have said how great his methods are for form cycle analysis. As is usually the case, when I actually test the ideas, they look pretty disappointing.

[Rick]

In case anyone's interested, my recent research into whether a recent form rating based on combined pace shows that I could increase my win % by about 1% but my ROI would stay the same (based on 440 races). So I probably won't be incorporating any factors like that soon.