Proposing a Profitable and Useful Horserace Betting Gizmo

[David Sklansky]

First, thank you Borodog. If it is not too hard, how bout some place, show, and exacta odds?

Second, my "what beats the rake" answer to PTB was misinterpreted. I should have written "WHAT beats the rake?". In other words I was asking for clarifacation on his question, not expressing doubt about beating it.

Third, the idea that parimutual odds defeats everybody is only true if everybody knows the true odds. If only a small percentage of people know them and they don't bet too early or too much there is plenty of money to be made. Owning this device will NOT help most people too much and might even hurt them since they will be inputting bad opinions.

Fourth, I don't think the SD should be anything calculated from past performances of that horse. It should depend on the specific circumstances. Perhaps you start with a known default SD for all horses of that category in races of that distance on that track and adjust from there.

Fifth, If the normal distribution is not the proper one, the gizmo will be changed to use that distribution. Actually you don't have to use a distribution at all. You could manually for each horse, assign probabilities to the hundredth of a second. More time consuming to be sure but still easily done.

Sixth and most important. The idea behind this idea is exacly analogous to the poker situation where an expert player is sure of the individual probability parameters connected to a decision (number of outs, chances he will call, chances he will check next round, etc, etc) but isn't sure how they mathematically combine. He miscalculates and plays it wrong.

I have sometimes been criticized when I show how to combine hypothetical assessments into a final decision. Critics ask how someone is supposed to know that there is "a 20% chance he is bluffing". But what does that have to do with the technique? Agreed that most people will screw up the decision because they mis estimated the parameters. But some will come to the wrong decision with good estimates followed by bad math. And those people can be helped.

Same goes for horseracing and this Sklansky Horserace Calculator (patented by Borodog Enterprises). Good handicappers first have to analyze each horse and then translate that analysis into the chances of winning. (That's simplistic since they also have to analyse other factors and consider exotic bets as well.) But just like in poker, the mere fact that you can perfectly predict the chances of a horse running a certain time does not mean that you can predict the chances he will finish in a certain spot. The math is just too complicated. But computers can do it in seconds and unlike poker you have plenty of time to make your decision.

How much this gizmo would help you depends on how well you know horses and how bad your mathematical intuition is. Obviously, to use it you would have to learn the basics of the normal distribution, if that is what is used, in order to input SDs. An interesting experiment would be to make a line based on your overall intuition and compare that to the line that the SHC comes up with based on your individual intuitions. Theoretically the SHC line should be better. If it wasn't you should reexamine your intuition regarding SDs.

Once it is clear that the SHC line is better than your intuitive line you would obviously use it to find overlays. Bet on horses or exotics where the SHC line is more probable than the tote board. How much of a discrepancy would be required is similar to the criteria you should have been using BEFORE you had the SHC. You must take into account the possibility the line drops, your own uncertainty about your ability to peg things perfectly, and your willingness to take close gambles.

If you are already a winning or close to winning handicapper, I would think the SHC will identify some good bets that you are not presently making and get you to avoid some bad bets that you are. If you are a so so handicapper, you might turn into a winning one and you might not. If you are an amateur, you won't admit it and you will buy the SHC anyway.

[mojobluesman]

David,

I'm not much of a poker player yet, but I am a long term winning horseplayer and have been for well over a decade.

You have a lot of useful insights in your post, but some of this has already been tested with relatively poor results.

You identified the "general problems", but they occur far more frequently and for a wider variety of reasons than I think you suspect.

Here are just a few:

1. Pace effects final time but it is more difficult to measure accurately than final time because of track configuration, changing wind direction and strength, rating by jockeys, the run up from the gate to the starting line for timing races varies, and other factors. This subject alone requires a book.

2. Horses change trainers and this often has a huge impact on performance.

3. Some tracks tend to be biased (not all the paths are equally fast or favor the same running style)

4. The competitive development and preferred running styles of the horses appear to impact their final time.

5. Ground loss impacts time

6. Horse's get into trouble, blocked, checked etc...

I could go on and on.

If you build your odds line primarily based on time, it will tend to identify theoretical overlays that actually aren't. That's because you wouldn't be incorporating these other factors into past performances or the potential for them being factors in today's race as well as the public is as a group.

There are lots of roads to heaven, but I can say with 100% certainty that there are some recurring inefficiencies in the pools that can be exploited regularly that reduce the impact of the track take significantly. With incremental handicapping insights it is possible to get a reasonable long term edge.

You can also get an edge via handicapping alone, but it generally takes a higher degree of selectivity.

It's a very rough game to beat for significant money. Those that do it (like me) would be far better off concentrating entirely on poker or the stock market etc.... It's just that most horseplayers enjoy the sport and sightly different intellectual challenge more.

Feel free to ask me anything you want.

[creedofhubris]

I wonder if you could generalize this principle to locate profitable bets on "erratic" teams and players.

[David Sklansky]

You are giving me the poker objection. And it is well taken. But it still doesn't change the fact that some bets are missed or made by pros because multi player contests with individual standard deviations sometimes results in counter intuitive probabilities. The expert horsebettor would shade his line anytime he felt the SHC spit out something surprising. How much he would shade it would depend on the specific circumstances you mentioned.

Even great poker players find it hard to intuitively believe many of the mathematically correct move ins playing no limit holdem. But they now do it anyway. I think this calculator would expose analogous situations, especially because you can instantly ask it about EVERY POSSIBLE BET.

Also the deluxe model will allow inputs that don't assume independence. In other words if two horses raced next to each other and ship in from another track you can specify one horse"s average time and say the second horse is .1 seconds behind it (with a certain EV).

The fact is that if you simply used published speed ratings and a generic one size fits all SD the SRC would likely come up with surprising and usable results from time to time.

[David Sklansky]

"If you build your odds line primarily based on time, it will tend to identify theoretical overlays that actually aren't. That's because you wouldn't be incorporating these other factors into past performances or the potential for them being factors in today's race as well as the public is as a group."

Te expert handicapper would try to incorporate those factors into he SHC input. Either the time or the SD or both.

[Borodog]

[ QUOTE ]
First, thank you Borodog. If it is not too hard, how bout some place, show, and exacta odds?

[/ QUOTE ]

David,

You are very welcome. Tomorrow morning. Off to bed.

[Borodog]

Horse 1 to win: 0.3043955
to place: 0.2737382
to show: 0.1999276

Horse 2 to win: 0.2730526
to place: 0.2072023
to show: 0.1731647

Horse 3 to win: 0.1187247
to place: 0.2277926
to show: 0.276182

Horse 4 to win: 0.0845282
to place: 0.1709041
to show: 0.2427186

Horse 5 to win: 0.219299
to place: 0.1203628
to show: 0.1080071

Exactas
Win\Place 1 2 3 4 5
1 -------------- 0.0947495 0.0930645 0.0663041 0.0502774
2 0.1126503 -------------- 0.0677799 0.0481196 0.0445028
3 0.0474221 0.0306293 -------------- 0.0253095 0.0153638
4 0.0312456 0.0201805 0.0228833 -------------- 0.0102188
5 0.0824202 0.0616430 0.0440649 0.0311709 --------------

10,000,000 races.

[Borodog]

Or, in odds to one against,

Horse 1 to win: 2.2851996
to place: 2.6531255
to show: 4.0018106

Horse 2 to win: 2.662298
to place: 3.8262012
to show: 4.774849

Horse 3 to win: 7.4228473
to place: 3.3899581
to show: 2.6208007

Horse 4 to win: 10.830372
to place: 4.851235
to show: 3.1199975

Horse 5 to win: 3.5599842
to place: 7.3082147
to show: 8.258651

Exactas
W\P 1 2 3 4 5
1 ------------- 9.5541460 9.7452350 14.082024 18.889652
2 7.8770290 ------------- 13.753636 19.781553 21.470497
3 20.087215 31.648478 ------------- 38.510857 64.088066
4 31.004507 48.552788 42.699990 ------------- 96.858850
5 11.132948 15.222442 21.693800 31.081203 -------------

[mojobluesman]

[ QUOTE ]
You are giving me the poker objection. And it is well taken. But it still doesn't change the fact that some bets are missed or made by pros because multi player contests with individual standard deviations sometimes results in counter intuitive probabilities.

[/ QUOTE ]

I think your points here and in the original post are very well taken.

To some degree, things like that "are" built into the odds, but I don't know if it is done very well. So if you could do it well, it would be useful.

For example, extra credit is typically given to horses that have "back races" that are good enough to win, even if their more recent form hasn't been as good. However, they are sharp enough to not give as much extra credit to the horse in the place and show pools. They know that some categories of horses are "all or nothing".

Again, these kinds of thing may not be built in properly and there may be many counter intuitive situations you could exploit. I am looking forward to your contributions.

[Borodog]

David,

Seeing as I just got the bill for the damage to my car, $2574, for a mere 10% of your profits from its use I can provide you with an executable version that will allow you to input horses, times, standard deviations, and the number of races you want run, that will produce a nice standardized output file similar to the format above. If I knew jack about horses I might take a crack at it myself, but alas, I don't bet sports.