Finding overlays in the exacta pool

[formula_2002]

I've been down loading the Twinspires tote board data for over a year, mulling through reams of data (actually data files of data).

I find that I can point out the overlays with 1 minuet to post, and on average, the return is better than expected.

That is in, in 10 races with winning overlays, a $100 distributed bet on all overlays in proportion to their value retuns $110.

The only problem!!
Not all races are won by an overlays!!

Based on my "overlay calculations" at 1 minuet to post time, only 40% of the races are won by overlays... :eek:

Forty percent is probably as good, if not better than the final win odds calculations for an overlay.

[pondman]

Forty percent is probably as good, if not better than the final win odds calculations for an overlay.

Not sure how you are defining overlay. To me it's all about the spread between my own line and the crowd's. I won't play unless I get 3 or 4 extra points on the nose. I've yet to believe the exacta is a good deal, at least for me. A tri can produce returns in the +5k range, if I key, all, all on a 25-1 spot and the favorite runs out. But my math says it about the same as on the nose overtime.

[formula_2002]

the overlay is based on the public's odds as determined in the win pool, with some adjustments.

[Some_One]

What I've been working on is trying to find races where I think it is predictable and I think the public is wrong in some fashion. In exotics, when the favs run out, it always multiplies the implied parlay of the payoff. I've been tracking this with a study in HSH as I'm using NP & BoW for my system.

So far so good here:

EXACTA RESULTS
Races: 252
Pays: 52
Pct: 20.6%
Total Return: 8,730
Total Bet: 5,304
Result: 3,426
$Net: $3.29
(Overall, the fav has a 25% win percentage, largely because I think I have a good understand of what type of races favs win)

Of course I need probably another 750 races to get a good sense of where I'm at.

[lamboguy]

i love it when someone uses the word "public". to me when someone uses that term, it infers that person using that word thinks he is smarter than the so called "public", and that the "public" always makes mistakes.

the "public" is the market. and to me there is no one entity that is smarter than the market. the market is alway's the final arbitrator.

on the other hand, if you want to tell me that the majority of people tend to make poor assumptions and decisions, i would agree with that statement. but even if you have a high proportion on people that are in a market, they are not the complete market, and even if that proportion is as high as 90%, they might not account for the highest proportion of money in a market.

[Turkoman]

The "public" is a pretty good handicapper. ;)

[formula_2002]

The "public" is a pretty good handicapper. ;)

Yes they are..except their sense of fair play sucks!!
betting into a 15% up to 18% win pool take!!
25% exacta pool take...
I study horse racing, but I play craps (1.14% take)
It make's me much less the fool :)

[formula_2002]

What I've been working on is trying to find races where I think it is predictable and I think the public is wrong in some fashion. In exotics, when the favs run out, it always multiplies the implied parlay of the payoff. I've been tracking this with a study in HSH as I'm using NP & BoW for my system.

So far so good here:

EXACTA RESULTS
Races: 252
Pays: 52
Pct: 20.6%
Total Return: 8,730
Total Bet: 5,304
Result: 3,426
$Net: $3.29
(Overall, the fav has a 25% win percentage, largely because I think I have a good understand of what type of races favs win)

Of course I need probably another 750 races to get a good sense of where I'm at.

Things I look at when evaluating my different systems.
If you are playing 1 exacta combination per race, what's the win pool $net on your top horse?
If playing more than 1 combination, what's your normalized return based on the expected value of the combinations. I do that for sinlge combination plays also.
What is the exacta payout compared to the public's win pool odds.
I'll do most any analysis to discredit what appears to be my winning system, and I've never been wrong!!! :)

[formula_2002]

What I've been working on is trying to find races where I think it is predictable and I think the public is wrong in some fashion. In exotics, when the favs run out, it always multiplies the implied parlay of the payoff. I've been tracking this with a study in HSH as I'm using NP & BoW for my system.

So far so good here:

EXACTA RESULTS
Races: 252
Pays: 52
Pct: 20.6%
Total Return: 8,730
Total Bet: 5,304
Result: 3,426
$Net: $3.29
(Overall, the fav has a 25% win percentage, largely because I think I have a good understand of what type of races favs win)

Of course I need probably another 750 races to get a good sense of where I'm at.

Things I look at when evaluating my different systems.
If you are playing 1 exacta combination per race, what's the win pool $net on your top horse?
If playing more than 1 combination, what's your normalized return based on the expected value of the combinations. I do that for sinlge combination plays also.
What is the exacta payout compared to the public's win pool odds.
I'll do most any analysis to discredit what appears to be my winning system, and I've never been wrong!!! :)

[jdhanover]

What I've been working on is trying to find races where I think it is predictable and I think the public is wrong in some fashion. In exotics, when the favs run out, it always multiplies the implied parlay of the payoff. I've been tracking this with a study in HSH as I'm using NP & BoW for my system.

So far so good here:

EXACTA RESULTS
Races: 252
Pays: 52
Pct: 20.6%
Total Return: 8,730
Total Bet: 5,304
Result: 3,426
$Net: $3.29
(Overall, the fav has a 25% win percentage, largely because I think I have a good understand of what type of races favs win)

Of course I need probably another 750 races to get a good sense of where I'm at.

Awesome!!! If you really are able to pick races where faves only win 25% of the time, it should translate to a huge advantage as your study shows. The whole key will be if that 25% stays or goes up to 'normal' (~37% I think) in the next 750 races.

In HSH you can also slice/dice to see what is working best bet wise. One thing to look at as well...what is the $net without the top 2 or 3 hits. I've had studies look good then to find it is driven by one or two very big hits - those big hits don't necessarily repeat hence I pull them out.

[Some_One]

Awesome!!! If you really are able to pick races where faves only win 25% of the time, it should translate to a huge advantage as your study shows. The whole key will be if that 25% stays or goes up to 'normal' (~37% I think) in the next 750 races.

In HSH you can also slice/dice to see what is working best bet wise. One thing to look at as well...what is the $net without the top 2 or 3 hits. I've had studies look good then to find it is driven by one or two very big hits - those big hits don't necessarily repeat hence I pull them out.

I agree, I wonder about the big hits are causing to my deviation, but I started this trying to profile in what type of races favs do great in, and then simply to skip them. I guess I am swinging for the fences, but I'll be honest and say the win betting roi is something like 10%, so the exotics are showing that when I'm right in finding a weak fav, usually I get the exactor too with my contenders and I get a good payoff. In testing I had some boring days during the week with something like only 6 races playable, so be it. When it comes to the $net, and this goes back to the odds line thread, I've simply seen it really doesn't matter that much when it comes to price horses, I guess it's only been useful in isolating weak low price horses.

[teddy]

I think at one time or another we have all tried to pick races where there is a false favorite with software. Usually on mine its when the chalk is my third or fourth pick from my software. Thus you have to watch each race to see if its a play against in exactas.

[teddy]

betting all overlaid exactas in races with bad chalks?? IF thats what you are saying it would be daunting to get that bet down in the last min.

[davew]

EXACTA RESULTS
Races: 252
Pays: 52
Pct: 20.6%
Total Return: 8,730
Total Bet: 5,304
Result: 3,426
$Net: $3.29



What are your largest 5 pays?
and if you lost those 5 races (10%), what would be your net?

[formula_2002]

06-15-2012
belmont r7
at 1 min to post, with $107,000 in the exacta pool, the 7,4 exacta was $12.00, it closed at $6.70.

Based on the final odds of the two horses, the fair exacta was $10.79.

Try making money with that..

[castaway01]

06-15-2012
belmont r7
at 1 min to post, with $107,000 in the exacta pool, the 7,4 exacta was $12.00, it closed at $6.70.

Based on the final odds of the two horses, the fair exacta was $10.79.

Try making money with that..

You should stick with craps...why bother with 10,000 more posts about how you can't win with racing if you blindly bet the same thing as the public?

[formula_2002]

an update.
1 minuet to post
top pick (most often the favorite)

top pick over all exacta
all over top pick exacta

bet size normalized in proportioned to size of overlay


100 races, usa flats, usa trotters, uk, saf
underlay winning exacta with the top pick (1st or 2nd), returned .73 roi
overlay winning exacta with the top pick (1st or 2nd), returned 1.00 roi

underlays won 37% of the races
overlays won 19% of the races.

[formula_2002]

156 race up date

1 minuet to post
top pick (most often the favorite)

top pick over all exacta
all over top pick exacta

bet size normalized in proportioned to size of overlay


156 races, usa flats, usa trotters, uk, saf
underlay winning exacta with the top pick (1st or 2nd), returned .75 roi
overlay winning exacta with the top pick (1st or 2nd), returned 1.03 roi
underlays won 36% of the races
overlays won 17% of the races.

[formula_2002]

here is a list of the winning overlay returns for $100 proportional bet

59.00
153.00
172.00
174.00
212.00
228.00
230.00
261.00
266.00
301.00
305.00
310.00
330.00
370.00
380.00
385.00
403.00
480.00
501.00
512.00
558.00
610.00
700.00
715.00
770.00
819.00
825.00
847.00
1027.00
1065.00
2138.00

[formula_2002]

165 races

overlay roi = 1.06
underlat roi =.075


165.00 95.00 0.75 1.06
# races <1 roi >1 roi
wins

[formula_2002]

a $4.05 dollar exacta may not seem much, but it was an overlay in my matrix. A $100 proportional bet returned $202. $2.02 on the dollar, compared to a 30 cents win pool dollar



Woodbine T-bred - Race 5
# Win Place Show
2 $2.60 $2.30 $2.10
6 . $3.70 $2.60
1 . . $2.30
Results: 01 | 02 | 03 | 04 | 05 | ALL
Wager Type Winning Numbers Paid
$1.00 Exacta 2-6 $4.05
$1.00 Trifecta 2-6-1 $8.90
$0.20 Superfecta 2-6-1-5 $6.03
$0.20 Pick-3 8/4/2 3 of 3 $10.27

[formula_2002]

My exacta matrix had the top two choices, in order, as the only overlay.
a $100 "proportional bet" (100%) returned $400!



Churchill Downs - Race 5
# Win Place Show
8 $3.80 $2.40 $2.20
3 . $2.60 $2.20
5 . . $3.40
Results: 01 | 02 | 03 | 04 | 05 | ALL
Wager Type Winning Numbers Paid
$2.00 Exacta 8-3 $8.00
$0.50 Trifecta 8-3-5 $6.50
$0.10 Superfecta 8-3-5-7 $2.36

[formula_2002]

204 race update
By the way, there is a play in every race!!

underlay roi .77
overlay roi, 1.04

underlay win % = 38
overlay win %= 21

overall roi = .91

[formula_2002]

today, I charted 23 races. a win bet on the top pick returned $41.2 for the $46 bet., a .89 roi

a proportional bet on the top pick exacta overlays, returned $2758 for the $2300 bet , a 1.18 roi.
A good day for the exacta, but more importantly, is how well the exacta out performed the win bet.

I'll track the comparison in the future.

[formula_2002]

SAF Arlington r8
this was the 1st wiiner in todays 1st four races.
$6.80 win mutual and a $472 exacta (FOR A $100 PROPORTIONAL BET).

$ return for the win bets was 6.80/8=.85 roi
while the exacta roi was $472/400 = 1.18 roi


sow's ear into silk purses? :)

[formula_2002]

update

the top pick, over all overlays, returned $23602 for 217, $100 proportional bets, for a 1.087 roi.


I've been using the same method for selecting the top pick in the selections forum for well over a year. I never looked into using it in an exacta until recently.
Overlay exacta's seem to be doing very well :cool:

it goes further back than this
http://www.paceadvantage.com/forum/showthread.php?t=89538

[formula_2002]

311 race update
there is a play in every race!!

underlay roi .77
overlay roi, .94

underlay win % = 38
overlay win %= 19

overall roi = .86
__________________

[formula_2002]

two big wins just put the overlay roi at 1.00

Also, I have a new way of determing the bet size for the underlay plays which could put them into the positive roi.

[formula_2002]

today's new underlay format returned a .96 roi in 42 races.
the overlays returned a 1.52 roi in 45 races

[formula_2002]

update:
in 82 races, the new underlay format returned a 1.04 roi
in 397 races, the overlays returned a 1.00 roi

[formula_2002]

update:
in 82 races, the new underlay format returned a 1.04 roi
in 397 races, the overlays returned a 1.00 roi

july 2 2012 update

in 207 races, the new underlay format returned a .94 roi and a win % of 35%
in 533 races, the overlays returned a .88 roi and a win % of 19%
I've recorded the win mutual for 352 of those races. winners returned a win pool roi of .86 and a win % of 32%.

[formula_2002]

another restructuring in the quest for overlay profits.

Plain vanilla.
No handicapping or fancy betting structures.
Just using the public's win pool and exacta pool probability to find the overlays at 1 minuet to post.

after 115 races.
expecting win probability for all overlays =44.86 %
actual probability = 60.87%

roi=1.0178%