Estimating Trifecta Dividends

[arkansasman]

Does anyone know of a good way to estimate trifecta dividends?

John

[steveb]

Does anyone know of a good way to estimate trifecta dividends?

John

john
isn't harville close enough to estimate trifecta dividends?
trifecta dividends have big variance anyway unless at short end of prices.
seems to be close enough in australia

may need another way to weigh your own investments, but to estimate public probability, i would have thought harville close enough.

steveb

[arkansasman]

Steve,

I have taken straight Harville in the past and divided 1-take by that. Sometimes it gets close, but very rarely. I was thinking that maybe I could estimate trifecta dividends by utilizing the Exacta Grid, but have not figured out a good way to do that.

John

[Pace Cap'n]

A rule of thumb I use is to require the odds to add up to 13...this will usually get a return of $150 'prox.

Have any DB guys ever attempted to correlate total odds to payoffs?

[steveb]

Steve,

I have taken straight Harville in the past and divided 1-take by that. Sometimes it gets close, but very rarely. I was thinking that maybe I could estimate trifecta dividends by utilizing the Exacta Grid, but have not figured out a good way to do that.

John

john

are you accounting for take on both sides of the equation?

if you use win the win approximates, then you need to take account of the win take in your calculations, as well as the trifecta take.

same would apply if using the exacta approximates, but personally think that would not work as good as using the win approximates.

works fine over here, but doubt you would ever get it really accurate on the combinations that don't have much action.
certainly works out pretty good in australia for the more favoured combinations.
however i am ignorant of how things work in usa, so i may well be way off track!

steve

[arkansasman]

A rule of thumb I use is to require the odds to add up to 13...this will usually get a return of $150 'prox.

Have any DB guys ever attempted to correlate total odds to payoffs?

Pace,

Can you explain this?

Thanks,

John

[rastajenk]

Steve Crist's blog (http://cristblog.drf.com/crist/2008/03/low-payoffs-at.html#comments) has a discussion about the Tampa Derby payoffs; the short version is that it was a unique occurrence with few comparisons, so who can say whether the payoffs were "short" or "generous." But even in races with many comparisons, I believe, there are so many variables that precise estimations are nearly impossible: field size, strong or soft favorites, two or more co-faves, pool size, etc.

The best way to guess trifecta payoffs is to read the charts everyday and just get a feel for it. Just how precise does it need to be, anyway?

[rokitman]

Did Crist actually write that a trifecta that he calculates should pay around $7000 and actually pays $439.20 is really not that suspicious at all? And part of his crack analysis is that if you take War Pass out of the pools, recalculate the exacta prices without him in the pools and multliple the exacta by 11 and the triple by 26 for "getting War Pass out" then the "payoffs don't seem so unusual or unreasonable." What the fock does that have to do with anything? What kind of math is that? Quite a high scientific standard he sets for himself.

He can start with getting his math correct before he moves on to his ridiculous extrapolations. I didn't do the triple but the exacta should have paid about $170, not $150, as he wrote, using the win pool percentages, which is predictive in a large pool. I have Excel spreadsheets that track that very thing and, believe me, you are wasting your time trying to find significant discrepencies, except in very small pools (smart people have software that automatically correct these discrepencies, as some of you sharpies know). Big Truck had a 10% probability of being in first (same as the win pool ) and Atoned a 9.1% probability of being in second to Big Truck. (Atoned's win pool percentage of 8.2% divided by 90% You remove Big Truck's win pool percentage for "the race for second.") That exacta's "correct" odds were 85:1. (.785/.0091 (.10X.091)=digital odds of 86.26. Tampa's exacta take is 21.5%).

OK, so some very big bettor was extremely confident that War Pass was going to run out of the exacta, for some reason, and really hammered the exacta. Same with him running out of the tri. Same with him running out of the super. Who are these people that are so confident that he won't even hit the board? A 1/20 shot. A horse with over $700,000 on him to show. I would really like to meet them.

I'm sure you guys are chocks full of stories of triples that you expected to pay around $400 but ended up paying $7000, right?! And I'm sure that it happens all the more, the bigger the pools. I'm sure that it is all very reasonable, and not so unusual, at all.

[rastajenk]

I don't think the people that cashed were confidant. More like wishful; gamblin'. Couldn't bet the 1-9 (which is what they saw on the monitors); but, being gamblers, couldn't pass, either. I met some of them, and none of them knew anything about a fever, or how he looked in the paddock, or if he was up against a tiring track, or any of the other post mortems out there. They just saw an opportunity and took a poke.

[PeterCottonTail]

Harville is known to be biased. Not so easy to do better. Here is one attempt:
http://www.financialmathematics.com/wiki/Code:gevrace

Cheers,
Peter

[ALostTexan]

I will answer this from an academic standpoint. At the University of Arizona Race Track Industry Program, they teach a trifecta estimation method that follows this equation:

Trifecta Payout=Win Payout x 2nd Place Payout x 3rd Show Payout

Now, this method is taught not as much for knowing an estimate for the tri payout, but instead we learn this based on the reasoning that one should know a decent estimation for the payout, and if the actual payout is 50% of the estimate or less then regulators should at least look into the pools for potential "wrongdoing". This is not to say that the payout couldn't be less than the estimation, but it gives regulators a point where bells should be ringing.

For what it's worth, the following are other estimates that are taught for the same reasons...

Exacta Payout = Win Payout x 2nd Place Payout

Quinella Payout = (Win Payout x 2nd Place Payout)/2

It amazes me at times that I am actually going to school to learn this stuff...

[ALostTexan]

You know, now that I re-read this thread, I think the original question was to estimate the payout before the race, so my previous post was pointless...

[gm10]

Harville is known to be biased. Not so easy to do better. Here is one attempt:
http://www.financialmathematics.com/wiki/Code:gevrace

Cheers,
Peter

hey this is pretty good
is there a version available for S-plus?

[Dave Schwartz]

Quinella Payout = (Win Payout x 2nd Place Payout)/2

This is the old standard "house quinella" from Nevada: Win times half the place.

It is also quite useable since win odds x .40 is a reasonable expectation of place.

Thus, a good estimate for quinella (before the race) would be:

WinPay1 x (((WinOdds2 x .4) x 2 +2) / 2)

Thus: if you had 4/1 and 3/1:

$10.00 x (((3.00 x .4) x 2 +2) / 2)
$10.00 x ((1.20 x 2 +2) / 2)
$10.00 x ((2.40 +2) / 2)
$10.00 x (4.40 / 2)
$10.00 x 2.20 = $22.00 = Quinella


Therefore, Exacta = $44.

Now, I do not know if the Exacta formula holds up or not, but that is the computation.

[raybo]

Steve Crist's blog (http://cristblog.drf.com/crist/2008/03/low-payoffs-at.html#comments) has a discussion about the Tampa Derby payoffs; the short version is that it was a unique occurrence with few comparisons, so who can say whether the payoffs were "short" or "generous." But even in races with many comparisons, I believe, there are so many variables that precise estimations are nearly impossible: field size, strong or soft favorites, two or more co-faves, pool size, etc.

The best way to guess trifecta payoffs is to read the charts everyday and just get a feel for it. Just how precise does it need to be, anyway?

Very astute observations! I'm a superfecta only player but the same is true in that pool as it is in the Tri pool.

You hit it on the head when you said there are many variables that have to be considered, ie: number of runners, pool size, price of favorites (top 4). Where or if they are ITM and in which positions they hit, whether or not the favorite wins (or more importantly, if he finishes in the top 4 spots, what the favorites odds are, etc., etc., etc.

As I stated in another thread recently, the best you can do is to estimate the minimum payout for your ticket (with your lowest odds horse winning, your 2nd lowest odds horse finishing 2nd, etc.) and then have in mind what minimum estimated payout you will accept, before placing the wager.

This minimum estimated payout can only be done through experience, at each individual track. There is no formula, that I am aware of, which can do this for you, however, you can set a few rules that will keep you out of trouble. When I estimate a minimum payout I am very conservative and then still hope some of my higher odds horses make it (the minimum payout hits just keep you in the game until the big ones hit, this is where you make the money). You can get my basic criteria with a search of my posts.