Racing Forum

completely agree

Posted By: Dunbar
Date: 16 May 02, 2:58 pm

In Response To: I told this to a horse winner, and he told me .... (ricky)

it would work in a perfect world...but there's one problem, its not a perfect world. There are some horses that if he doesn't win the race, there's no guarantee that he came in 2nd or 3rd, etc. Meaning that there are some horses that are basically all or nothing, they don't follow a typical distribution that we would all expect and which the formula you listed above assumes. Thus it is very possible for a horse to have win and place odds not really correspond as much to each other as the math guys might think. He said, its true that many horses will follow this, but there are some that may not. So you can not blindly use the formula on every horse, you still must know the horses to know if they have "skew" in them.

This is correct. There are certain all-or-nothing horses whose performance will not be well-predicted by the Harville calculation. I believe that in general, horses do not come in 2nd quite as often as predicted by the Harville model.

That being said, there is nothing simple that will give as good an estimate of fair exacta odds. I have seen more complicated variations of the Harville equation that may be more accurate. One of these was in an article by Benter that described the method his group used in Hong Kong to beat the Quinella there. (I don't know if I can dig up that article.) The article had a graph of expected profits (I believe from bets they made) comparing how they did with their formula to how they would have done using the Harville formula. Their refined formula out-performed the Harville formula significantly, but the Harville formula itself would have made significant money there, too.

For the purposes we are discussing here, namely to get an idea of the fair exacta odds once we have estimated the fair win odds, the Harville formula should be reasonably good. Any refinements would generally raise the threshold for betting. So, if you want to be conservative, you can add 10-15% to the numbers in my table. Or, for example, if you are of the opinion that War Emblem will either finish 1st or out of the money, then you would not want to use the numbers in the table at all for exactas with War Emblem 2nd. (of course, in that event, you should not be betting an exacta with War Emblem 2nd anyway).

--Dunbar

Messages In This Thread

Exactas--How To Calculate Fair Exacta Odds -- Dunbar -- 16 May 02, 11:00 am

Great Post, Dunbar! -- Jesse F. Knight -- 16 May 02, 1:06 pm

Absolutely- superb post-- -- D -- 16 May 02, 1:48 pm

I told this to a horse winner, and he told me .... -- ricky -- 16 May 02, 2:13 pm

completely agree -- Dunbar -- 16 May 02, 2:58 pm

Articles by Harville, Ziemba, Benter, etc -- 666 -- 17 May 02, 12:04 am

Ziemba Book -- Jesse F. Knight -- 17 May 02, 11:53 am

Good luck -- StevieY -- 17 May 02, 12:22 pm

I have one -- ricky -- 17 May 02, 12:30 pm

No, you should sell it to me:) (nt) -- StevieY -- 17 May 02, 12:35 pm

auction -- ricky -- 17 May 02, 12:53 pm

if anyone cares, its out on ebay (nt) -- ricky -- 22 May 02, 11:26 am

I ordered one from Amazon.com today -- Dunbar -- 17 May 02, 4:47 pm

I tried that twice... -- StevieY -- 17 May 02, 4:50 pm

Racing Forum is maintained by Pi Yee Press