Am I falling into a trap?

[Ray2000]

Here's the setup

I did a logistic regression analysis on several race factors using about 15000 races from 2013. Derived the coefs and intercepts for each of 20 different tracks and used them in a Computer program for making Win bets. It had an overall performance of 40% Strike, 0.95 ROI. I back tested it on 15000 races from 2012 and have now completed another 15000 from 2014. Very similar results for both testing years.

Here's the trap

If I look at a subset of ~ 5000 races (2012-2014) where the win payout price is >$4.00 but <$8.00 and test my program on just those races then the ROI performance my win selections jumps dramatically into a range of ~ 45% Strike, 1.20 ROI.

OF course, I realize that I won't know the winner's price at the time of making the bet... but if I limit my bets by conditional betting set at odds > 1.0 AND < 3.0 (at-1 MTP), can I expect the jump in ROI performance?

I have an un-easy feeling that my logic isn't sound and I'm mixing apples with oranges here.

Any opinions?

[ReplayRandall]

Originally Posted by Ray2000

Here's the setup

I did a logistic regression analysis on several race factors using about 15000 races from 2013. Derived the coefs and intercepts for each of 20 different tracks and used them in a Computer program for making Win bets. It had an overall performance of 40% Strike, 0.95 ROI. I back tested it on 15000 races from 2012 and have now completed another 15000 from 2014. Very similar results for both testing years.

Here's the trap

If I look at a subset of ~ 5000 races (2012-2014) where the win payout price is >$4.00 but <$8.00 and test my program on just those races then the ROI performance my win selections jumps dramatically into a range of ~ 45% Strike, 1.20 ROI.

OF course, I realize that I won't know the winner's price at the time of making the bet... but if I limit my bets by conditional betting set at odds > 1.0 AND < 3.0 (at-1 MTP), can I expect the jump in ROI performance?

I have an un-easy feeling that my logic isn't sound and I'm mixing apples with oranges here.

Any opinions?

Now find just the exact opposite subset of the worst ROI produced over 5000 races. Now, hold out those worst races and the best producing ROI races(total of 10K). Once completing that, you should have a total of 35K races left out of the starting figure of 45K. Now tell me what the ROI of that set of races is.....

[osophy_junkie]

Try integrating the "final known odds" into your regresssion alrgorithem. This is the documented way of doing it.

[JustRalph]

Field size............ don't forget field size........

[Ray2000]

ReplayRandall

The worst 5000 would be the Longshots > 10/1, ... that subset only returned 2000 on 10000 bet

When I posted this, I picked a subset range $4-$8 to demonstrate the problem/confusion in an example. It's not exactly the best overall set, so I could do the math you suggest but maybe I should elaborate first.

The attached image shows a plot of bank balance for 2$ bets on all 44,281 bets sorted in order of increasing mutuel return prices.
The region between the blue lines is the range of $4 to $8 prices and has the best (steepest) return. (the X axis is the bet number). The region I used is about 15000 bets spread over 3 years.

I guess the approach you mentioned would be more meaningful if I'd just use a subset < $8.00. I'll take a look at it that way.

osophy_junkie

I did use 'Final known odds' in my early runs but it washed out all other factors except Morning Line. Maybe I should revisit.

JustRalph

Always a good point, it gets reflected in the returns.

[whodoyoulike]

This makes sense on the surface to me. By considering the >4 and <8, you're looking at a subset which has a higher hit rate than your total population.

Aren't you a harness handicapper? You have to realize I don't understand harness racing.

Because in Tbred's, I've noticed that <$8.01 odds win something like 70 - 80% (from memory) of the time in my databases. Try a random sample of your total database population for >2 and <8.01 and see what your results will be. Also, try it on your conditions of >1 and <3 and compare results.

But, I don't bet based only on odds. I bet on who I think is the best horse given my perceived pace scenario and only if the odds are acceptable because I will bet on horses >8.

I don't bet horses <2 but, I wouldn't bet on me always following this statement either.

[Ray2000]

whodoyoulike

Yes I'm a Harness bettor but the problem I'm trying to wrap my brain around could be for trots, runners or puppies.

I noticed a mistake in the OP. It should be 15000 not 5000 in the 3 year sector between Even money and 3/1s .....($4-$8)

[whodoyoulike]

Originally Posted by Ray2000

whodoyoulike

Yes I'm a Harness bettor but the problem I'm trying to wrap my brain around could be for trots, runners or puppies.

I noticed a mistake in the OP. It should be 15000 not 5000 in the 3 year sector between Even money and 3/1s .....($4-$8)

Are there pace scenarios in harness? Because, I really don't know.

Again, try a random sampling and see what you get. I think betting strictly on odds is dangerous and really doesn't make any sense to me or provide me with enough enjoyment for the game.

And, I'm not attempting to discourage you. People have to figure things out for themselves.

[traynor]

Originally Posted by Ray2000

Here's the setup

I did a logistic regression analysis on several race factors using about 15000 races from 2013. Derived the coefs and intercepts for each of 20 different tracks and used them in a Computer program for making Win bets. It had an overall performance of 40% Strike, 0.95 ROI. I back tested it on 15000 races from 2012 and have now completed another 15000 from 2014. Very similar results for both testing years.

Here's the trap

If I look at a subset of ~ 5000 races (2012-2014) where the win payout price is >$4.00 but <$8.00 and test my program on just those races then the ROI performance my win selections jumps dramatically into a range of ~ 45% Strike, 1.20 ROI.

OF course, I realize that I won't know the winner's price at the time of making the bet... but if I limit my bets by conditional betting set at odds > 1.0 AND < 3.0 (at-1 MTP), can I expect the jump in ROI performance?

I have an un-easy feeling that my logic isn't sound and I'm mixing apples with oranges here.

Any opinions?

Reality, not a trap. Possibly the result of computer-assisted wagering. I have found (and written about) similar results.

There is not a "smooth progression" in odds ==> win rate. Specifically, there is a tendency for very short odds entries to return way less (and to win way less) than would seem "reasonable." I first wrote about noticing this trend at Fraser Downs--in direct opposition to the "whale fantasies" of others, a heavy influx of late money is very often on losing entries. You may recall I wrote about a "signature" of computer-assisted wagering that kept odds close to even throughout the wagering process, rarely dipping below 1/1 or 4/5, and just as rarely going off at 3/5 or less.

It is not so much the odds at the last minute that matters as the process leading up to it. Much of the last minute dump is from "unsophisticated bettors pursuing what they erroneously believe to be value wagers."

What you are seeing is reality--not conjecture. And--as I pointed out elsewhere--closely monitoring the exacta pools will provide far better predictors than the win pool (which are highly predictive if you know what to look for).

The key point is to avoid the fantasy that some seem to cherish--of "whales" making millions by "churning for rebates." Believing in that particular fantasy can be costly (if it goes) beyond the talking stage.

[traynor]

You may also recall that I mentioned "discovering" (with Katherine Jung's assistance and encouragement) that as soon as I culled my selections going to post at odds OVER 3/1 at Fraser Downs (the $8.00 top end you mention above), my ROI took a major leap up. That is not "just at Fraser Downs" nor is it "limited to harness races."

It is every bit as applicable at thoroughbred races. "Seeking value wagers" could be as foolish an endeavor as betting random numbers in the hope of divine intervention.

Not my "theory." Reality. I just use it for leverage.

[MJC922]

Not sure about harness but in the thoroughbreds with a win% that high combined with trying to confine your bets to such a relatively high odds range I suspect you will have 40% of the horses going off below the range even if betting very late. If you can factor in the drop somehow it might be possible. When it comes to real world betting I'm increasingly dissatisfied with minimum odds requirements and have moved away from them of late.

[traynor]

Originally Posted by MJC922

Not sure about harness but in the thoroughbreds with a win% that high combined with trying to confine your bets to such a relatively high odds range I suspect you will have 40% of the horses going off below the range even if betting very late. If you can factor in the drop somehow it might be possible. When it comes to real world betting I'm increasingly dissatisfied with minimum odds requirements and have moved away from them of late.

The maximum odds are a far more interesting (and profitable) area to explore. Especially if one can avoid the don Quixote syndrome of believing one is cleverly "out-thinking" the expert bettors.

It is relatively trivial to establish that one's ROI can be increased with almost any consistent selection process by capping odds. In other words, the complete opposite of what "value bettors" do, and the complete opposite of what those who fail to factor for outliers in short samples do.

[traynor]

Originally Posted by MJC922

Not sure about harness but in the thoroughbreds with a win% that high combined with trying to confine your bets to such a relatively high odds range I suspect you will have 40% of the horses going off below the range even if betting very late. If you can factor in the drop somehow it might be possible. When it comes to real world betting I'm increasingly dissatisfied with minimum odds requirements and have moved away from them of late.

A big saver is the conditional wagering at 0 minutes to post. Lots of the entries that take a dip into the 1/5, 2/5 range after the off (from 1/1 or 6/5 earlier) are losers. If one understands the process, many of the odds drops are on losers chased by those seeking "sure things that the experts/insiders/smart money must think is going to win." It is as often that the odds on the best choice go UP after the off, because the online bettors chasing "value" bet on other entries.

Few things satisfy the soul quite as much as making a sizable wager on an entry at even money, seeing a last minute dump on other entries perceived to offer "better value," and watching the horse you wagered on romp to an easy victory at 9/5.

[MJC922]

Originally Posted by traynor

The maximum odds are a far more interesting (and profitable) area to explore. Especially if one can avoid the don Quixote syndrome of believing one is cleverly "out-thinking" the expert bettors.

It is relatively trivial to establish that one's ROI can be increased with almost any consistent selection process by capping odds. In other words, the complete opposite of what "value bettors" do, and the complete opposite of what those who fail to factor for outliers in short samples do.

The model that is worthwhile to me has to have inherent value and as such I suspect it should make some money without considering the odds -- not much money but some several pct. Certainly such a model will make more money at certain odds ranges but leveraging that fact in anything but a large pool is well beyond my current capabilities. I agree with you on value capping.

[ReplayRandall]

Originally Posted by Ray2000

The worst 5000 would be the Longshots > 10/1, ... that subset only returned 2000 on 10000 bet.

If what you wrote is true, if you never bet a play greater than 10-1, I have 40K races bet with a return of 83,500, thus giving a ROI of 1.04375.