Racing Forum

Way off, IMO

Posted By: Dunbar
Date: 16 May 02, 7:51 am

Gary, I believe you are missing the point on this in a big way. This has nothing to do with whether you are betting an Exacta box or a straight exacta. It has to do with how you estimate fair exacta odds if you know the fair win odds. And you are, IMO, severely underestimating the fair exacta payoffs.

Dunbar wrote:
"It doesn't matter if you call them "cutoff" or "fair odds", in either case they are too low. The numbers you gave are below BOTH ways in the exacta box, sometimes substantially, as I'll show below."

The Pro answered:
I disagree, and, I think, so would Beyer, Cramer, Mitchell, Quinn, and others.... In the case of an exacta box.

I would be very surprised if Beyer supported the way you are calculating fair exacta odds. Please quote me from one of his books where he suggests this. (or any of the other guys, for that matter)

The "fair odds" I posted were first probabilities before being converted to "fair odds" there is no need to again convert them to probabilities for determining "fair odds" on exacta boxes.
It has always been acceptable to me and to most others, in the case of exacta boxes to use "fair odds" x "fair odds" x 2 as a formula to determine the minimum you would accept in return for an exacta box.

Again, I would like to see a quote on this from a mathematically competent author.

It would not be acceptable to use any morning line, or tote odds, but as long as the odds have been converted to probabilities, as mine have been, "fair odds" x "fair odds" x 2 for exacta boxes is I believe more realistic than the Harville formula for the everyday player boxing exactas.

I have no idea why you believe this. Multiplying the win odds of 2 horses makes absolutely no sense, because both horses cannot win the race. I think you don’t understand the rationale behind the Harville formula, and I will go through that in a separate post.

For example you have two horses:

Horse A: "fair odds"= 2-1 probability= .3333

Horse B: "fair odds"= 4-1 probability= .2000

Exacta box: "fair odds" = 2 x 4 x 2 = $10 or 4-1

This is a gross underestimate of the fair payoff. The fair payoff should, in fact, be AT LEAST $22. (2-1,4-1 is $22; 4-1,2-1 is $26). That means that your cutoff has more than a 50% negative ev on both sides of the exacta box.

If you have made a good line, and narrowed your selections to two or three, you must have confidence your low odds/high probability horse will most of the time beat the higher odds/lower probability horse. In the above example I would accept 4-1 for my 2-1 horse winning. and if I have two horse listed with low odds/high probability as above, I'll be right just about the same number of times the probability suggests, which means long run profits.

This is a major error. Taking 4-1 for this exacta is betting with worse than a negative 50% expectation. The fair odds are at least 10-1.

Not optimal profits, but there is nothing optimal about making public selections 12-24 hours prior to post.

You are at a major disadvantage making picks so early. But using your own odds cutoff as a bet-decider for win bets (and by inference, exacta bets) would eliminate a lot of this disadvantage.

If I were going to list possible straight exacta wagers, I would use the Harville formula, but for exacta boxes I don't think the public will go for the extreme high odds Harville demands.

These are not “extreme high odds”. These are the odds that are necessarily to justify an exacta bet based on a good estimate of the fair win odds (or fair win probability) of the 2 horses comprising the exacta)

Nor do I think they care, so why go to the trouble. For public selections your talking mostly flat wagers for the vast majority, that's why I list them as I do.

I think the public wants to know that they are making a bet with some theoretical ev. Again, the point isn’t that you make different cutoffs for both sides of an exacta box. But your fair odds cutoff should AT LEAST be as high as one side of the exacta box. The way you are doing it, you are recommending a cutoff that is 25-50% below the real fair odds of EITHER side.

Having said all the above, which I know you disagree with, I'm sure others do also, and I want to at all costs avoid any controversy with my selections for the public on this site. So, I'll either not list any exacta wagers, or not list my version of exacta box "fair odds." That's why I didn't include them originally.

Well, I’d naturally prefer that I convince you to change the way you are looking at fair exacta odds, both for your own sake and for those on this page. But I don’t have huge amounts of time to do that. IMO, what you are posting now is so far off that you are better off posting no exacta odds.

We may disagree on this point, but I do always appreciate your comments, you are much more advanced with your handicapping than most will ever be. We can all learn from you knowledge and experience.

Thanks for the kind words. Have to disagree yet again, though. :>)

Gary

... Isn't the basis for the Harville formula to use tote odds, or other handicappers lines that are out of whack and convert to probabilities, then back to true "fair odds," rather than a line that has already been converted and represents true "fair odds?"

No, that’s not the “basis for the Harville formula”. What you describe is what Ziemba and Hausch did in their 2 books. Ziemba and Hausch use the tote board to estimate the fair win odds. (as opposed to doing their own handicapping) Then they plug these fair win odds estimates into the Harville formula to look for overlays in exactas, quinellas, place bets and show bets. It is always an estimate of fair win odds that goes into the Harville formula, whether obtained by your own handicapping or by some other method.

The rationale behind the Harville formula (and the formula itself) is quite simple and reasonable. I’ll post about it separately.

--Dunbar

Messages In This Thread

Way off, IMO -- Dunbar -- 16 May 02, 7:51 am

small correction -- Dunbar -- 16 May 02, 10:28 am

Racing Forum is maintained by Pi Yee Press