Exacta Overlays?

[Dutch]

I remember reading (Dick Mitchell, I think) about someone that played Exacta Overlays. I used to have a formula, and American Turf sells a slide rule that works it out. You use the win odds of two horses, and it tells you what the exacta SHOULD pay. You are then supposed to play only exactas that will pay more than they should. Or, you can look for smart money, if you believe in that. I used to test this, only on winning combinations. I checked races, and it seemed like the exactas never paid more than they should. So my conclusion was, if I played overlaid exactas, I would never hit any winning combos, so it didn't matter if they were overlaid or not.
It also bothered me that the assumption was that you could predict the exacta's payoff on the to-wijn odds, which doesn't really make sense. The Place horse doesn't have to win to be in the exacta.
I liked the old Dr. Z formula better, where you used the win/place/show pools to detemine when a place or show horse would pay more than it should. Only problem is, now with all the simulcasting, the pools aren't accurate enough to tell what will really happen. Another good theory goes down in reality.

[kenwoodallpromos]

Do you see any difference in toteboard overlays vs. your own value overlays for exactas?

[ceejay]

You might try:
http://www.turfpedia.com/playing_the_races/exacta_betting.html

[Dutch]

Well the point was, not YOUR value estimate, but an overlay based soley on the win odds (actual odds).
So... the chart from the link is exactly what I meant (although I've seen formula's for programmed calculators and slide rules - but it's the same idea). The page does say that they are frequently underlays, which is also what I had found from my experience. In other words, from the chart - a 2-1/ 3-1 exacta combo should pay $34. I found it might pay 28 to 30. It was extremely hard to find an overlay based on this. I felt this was because Exactas are overplayed in most likely combinations, by unknoledable players trying to hit something. Most of these plays may have been part of combinations.
A similar but ossible useless situration exists in some dog tracks. No one plays W/P/S - they are all trying for the big tri- or superfecta hit. So the pools are so small, a decent bets pounds the odds. And there is so much stupid money in the big pools, they don't pay fair either. The only good hit is when there is an unpopular choice, like a longshot in the show on a tri - one one race of a triple paying overly high. Makes it hard to make money.
I played with a guy once that made $30K playing weird superfecto combos. I also had to loan him money to finish the day - there was no logic to his method. When he hit, he hit big, but there was no way to do it with any consisentency.
Another point I made, was Is using the win odds (or your estimate of winning probability) even useful when trying to find a place horse for an exacta?

[andicap]

here's a question I have for Barry Meadow and others who post those exacta grids for getting "Value" on exactas?

When you say horses on your odds line are 2-1 and 5-1 you should get $XX back, are you talking about win odds? Because the odds on a horse to win are far different than its ability to place?
There are many horses who are 2-1 in the win pool on the board that I might give a lesser chance than that to win the race, but not necessarily to finish 2nd. Often I find horses have a much greater chance of finishing 2nd than winning. How do you incorporate that into these exacta "fair odds" formulas?

And I defy anyone to come up with an odds line for finishing 2nd.

This thought occured to me at Saratoga on Wednesday when a 4-1 shot beat a 9-5 Dutrow horse for a $27 exacta. I didn't like the 9-5 horse to win, but it was certainly a huge contender to place. So in my hypothetical odds line, the 4-1 was probably about right and the 9-5 horse was more like 4-1 to win.
$27 was probably an underlay in that scenario, but I thought the 9-5 horse had a great chance to finish 2nd (some horses who don't win are more likely to be off the board than be second -- e.g., "E" types who don't get the lead).

In that context, $27 may not have been a bad price.

[sjk]

andicap,

My computer does a separate place odds line and calculates expected exactas using win probabilities on top and place probabilities on the bottom. I would recommend it.

As you say, there are clear differences between the lines. As I think I have posted before, favorites and speed horses have lower odds in the win line and longshots and closers lower odds in the place line.

[formula_2002]

Andy its the win pool odds of the top horse x the "adjusted" win pool odds of the place horse. The odds of the place horse are equalized less the win horse odds.
The win pool odds % must also be adjusted for the win pool take-out.

I used a live tote board and gathered several thousand win pool and exacata pool pay outs. The data seems to support the formula.

It is also true that more post time exacta overlays have the shorter priced horse under the longer priced horse.

It is also true that the win pool odds tend to approach the calculated "win" odds of the exacta pool.

In a $2 exacta board, sum all the combinations (2/odds) for any one horse on top of "all" and you will get a win % for that horse based on the exacta pool payouts.


The win pool odds tend to approach the exacta pool win%.

With a minuet or so to post time if a horse is 1-1 in the win pool and 3/5 in the exacta pool, the final odds trend downward to 3/5.

if a horse is 2-1 horse in the win pool and 4-1 in the exacta pool, the final odds trend to the 4-1. If it doesn't, and goes to less then 2-1, that's due to late win pool money.

Joe M

[John]

Dutch,

This is just my way:

I have looked at countless exacta races. if you take the place payoff and multiply it by the winners payoff it is pretty close to the exacta payoff. [ your always in the ballpark] So if one of my horses is 3-1 and the other 5-1 I figure the winner will pay $8.00 and the the other around $4.00 to place 4X8=32. If the monitor says better than $32.00. I'll bet it.

I am sure that BillW or someone else will have something on the same line as this.

Hope this helps.

John

[JackS]

John- I think your right. (3/1x5/1)=3x5=15/1 or 30/2. These would be fair odds if it weren't for track take. This payoff would be close to fair in a Qunella since the ability of the 2nd place horse to finish 2nd is Increased by the first placed horse winning. In this case it's 25% (3/1). 25% could equalize the take. If reversing the two horses, the differences becomes the difference between 25%(3/1)and 17%(5/1) which would equal 8%, The 8% represents about 1/3rd of 25% ,so the amount of bonus required for the less favored horse on top should be about one third better than 30/2 or in other words 40/2.
Double check my figures but hope I'm close.

[John]

Jacks

Very Good, you explain it better than I