how often do two of the top three horses come in 1st and 2nd

[formula_2002]

over a large number of races, say 500, how often do two of the top three horses come in 1st and 2nd ?



appreciate your answer.

[davew]

top three horses by what criteria? how many horses in race? class level? surface? ....

[formula_2002]

top three horses by what criteria? how many horses in race? class level? surface? ....
none

just top three over about 500 races

[Robert Fischer]

smartass answer = Often enough to produce a stat that looks interesting, but not often enough to overcome the fact each individual race is independent in terms of how the top 3 horses will fare.


my guess = 25%?


if you define 'Top horses' as 'top 3 favorites' , someone on the board can probably give you that answer.

[formula_2002]

smartass answer = Often enough to produce a stat that looks interesting, but not often enough to overcome the fact each individual race is independent in terms of how the top 3 horses will fare.


my guess = 25%?


if you define 'Top horses' as 'top 3 favorites' , someone on the board can probably give you that answer.

yes "top three public's choice"

[lansdale]

over a large number of races, say 500, how often do two of the top three horses come in 1st and 2nd ?



appreciate your answer.

I would think this has increased in the past few years, but guessing ca. 45%.

[ReplayRandall]

I would think this has increased in the past few years, but guessing ca. 45%.

If the figure 45% is close to accurate, the payoff point average for profit would be at $7.40 for the $1 exacta....

[tanner12oz]

over a large number of races, say 500, how often do two of the top three horses come in 1st and 2nd ?



appreciate your answer.

when I bet them never

[Stillriledup]

none

just top three over about 500 races

Quite often the top 3 choices won't be decided until its too late to act as the last possible flash of money (and it's a large amount) comes in after the race has started. (Comes in before race starts but you don't see it till after race starts)

[AndyC]

What is the point in obtaining the information?

[Robert Fischer]

I would think this has increased in the past few years, but guessing ca. 45%.

Yea, 45% sounds more reasonable than 25%.

Someone should have the stats :)

[formula_2002]

If the figure 45% is close to accurate, the payoff point average for profit would be at $7.40 for the $1 exacta....


with a 25% exacta take out, the min $1 payout should be,

1/(.45*.75) = $2.96

[AndyC]

with a 25% exacta take out, the min $1 payout should be,

1/(.45*.75) = $2.96

I, too, would go with the $7.40. The bet is $6.

[lansdale]

with a 25% exacta take out, the min $1 payout should be,

1/(.45*.75) = $2.96

Hi Formula-

A guy who went to Tech and has PhD shouldn't be accepting my 45% figure so easily. Hope others will check this, but when I went back and checked my figure, I realized I had doubled the outcomes, so the final figure should be ca. .225 - Robert F. was closer than myself on this.

So, to plug this in: 1/(.225*.75) = $5.92 for $1 payout/.25 takeout.

Since we're more likely to see this, the average $2 payout would be $13.84.

[lansdale]

Hi Formula-

A guy who went to Tech and has PhD shouldn't be accepting my 45% figure so easily. Hope others will check this, but when I went back and checked my figure, I realized I had doubled the outcomes, so the final figure should be ca. .225 - Robert F. was closer than myself on this.

So, to plug this in: 1/(.225*.75) = $5.92 for $1 payout/.25 takeout.

Since we're more likely to see this, the average $2 payout would be $13.84.

Wrong again - at least based on the figures I'm using - 1. .36/.18 - W/P
2. .25/.125 - W/P 3. .15/.075 W/P. I'm sure these values aren't exact, but are close to NA means.

Revised figure: 1/(.175*.75) = $7.63 for $1 payout/.25 takeout.

$2 payout: $16.63

Not sure what point Formula trying to make here - but wondering whether low odds exacta bets might have a compounding effect on the favorite/longshot bias.

[lansdale]

Wrong again - at least based on the figures I'm using - 1. .36/.18 - W/P
2. .25/.125 - W/P 3. .15/.075 W/P. I'm sure these values aren't exact, but are close to NA means.

Revised figure: 1/(.175*.75) = $7.63 for $1 payout/.25 takeout.

$2 payout: $16.63

Not sure what point Formula trying to make here - but wondering whether low odds exacta bets might have a compounding effect on the favorite/longshot bias.

Corrected $2 payout: $17.26

[Robert Fischer]

estimate win slot = about 1) 35% 2) 21% 3) 16% roughly 72% or so on average?

If place slot is similar (remember we are talking exactly 2nd, not place bet payout) 35% 21% 16%

.72 x .72 = 52%


not really in a puzzle solving frame of mind, but that is how I came to agree with about 50%

or if the place slot is a bit lower (all or nothing syndrome) , then maybe it's closer to 40% or or so...

someone should probably have these stats :eek:

[AndyC]

Corrected $2 payout: $17.26

Corrected payoff for what? To break even?

[lansdale]

estimate win slot = about 1) 35% 2) 21% 3) 16% roughly 72% or so on average?

If place slot is similar (remember we are talking exactly 2nd, not place bet payout) 35% 21% 16%

.72 x .72 = 52%


not really in a puzzle solving frame of mind, but that is how I came to agree with about 50%

or if the place slot is a bit lower (all or nothing syndrome) , then maybe it's closer to 40% or or so...

someone should probably have these stats :eek:

Looks like whoever has this number doesn't want to post it - maybe for good reason. Where is Jeff P.? Re your figures here - don't see how this can be right - is this not a problem in combinatorics? - something I never studied. You add the products of the six combinations. Your assumption that place probability is only slightly lower than win is not correct. In general, place and show frequencies each tend to be almost exactly half of the win frequency. You're making the same mistake that I did, at first in doubling the outcomes.

As you say, though, these (or more accurate) figures are only based on actual rank performance frequency, irrelevant of exacta odds. My guess is that the actual mean payoff for an $1 exacta of the three top odds horses is closer to $5.50 and $11.00 for $2.

[lansdale]

Corrected payoff for what? To break even?

For a $12 3-horse exacta box bet. Hope you realize we're pulling these numbers out of the air - but would be interesting to know what the real number is.

[AndyC]

For a $12 3-horse exacta box bet. Hope you realize we're pulling these numbers out of the air - but would be interesting to know what the real number is.

Using your probability of hitting, the odds of hitting one of the exactas in the box is 4.71/1, computed .825/.175. So to break even your $2 winning tickets need to average $68.52. ( $12 X 4.71 + $12)

[MONEY]

Today's tracks

Finger Lakes
Hawthorne
Mahoning Valley
Parx Racing
Portland Meadows
Turf Paradise
Zia Park

Today there were 63 thoroughbred races.

The off time favorite won 30 times/placed 12 times.

The 2nd favorite won 11 times/placed 12 times.

The 2 favorites combined to make the exacta 13 times.

Neither of the favorites finished 1st or 2nd 11 times.

Today the 2 favorites finished 1st or 2nd 81% of the time.

Yesterday I checked how many times the 2 chalks finished in one of the
top 2 top spots & I remember it being 78% of the time.

[thaskalos]

Today's tracks

Finger Lakes
Hawthorne
Mahoning Valley
Parx Racing
Portland Meadows
Turf Paradise
Zia Park

Today there were 63 thoroughbred races.

The off time favorite won 30 times/placed 12 times.

The 2nd favorite won 11 times/placed 12 times.

The 2 favorites combined to make the exacta 13 times.

Neither of the favorites finished 1st or 2nd 11 times.

Today the 2 favorites finished 1st or 2nd 81% of the time.

Yesterday I checked how many times the 2 chalks finished in one of the
top 2 top spots & I remember it being 78% of the time.

Yes...but you have not addressed this thread's original question. :)

[formula_2002]

CONSIDER THIS.
IF 33% OF THE FAVORITES WIN, THEN 66% OF ALL REMAINING HORSES PLACE.

IF 66% OF THE REMAINING HORSES WIN,
THEN FAVORITE PLACES 33% OF THE TIME.
THEREFOR THE TOP THREE HORSES MUST COMBINE FOR AN EXACTA PAYOUT LESS THAN 32.66% OF THE TIME.

.33*.66*.75 =.1633
.66*.33*.75= .1633
TOTAL .3266
THATS FAVORITE OVER ALL + ALL OVER FAVORITE

[formula_2002]

Today's tracks

Finger Lakes
Hawthorne
Mahoning Valley
Parx Racing
Portland Meadows
Turf Paradise
Zia Park

Today there were 63 thoroughbred races.

The off time favorite won 30 times/placed 12 times.

The 2nd favorite won 11 times/placed 12 times.

The 2 favorites combined to make the exacta 13 times.

Neither of the favorites finished 1st or 2nd 11 times.

Today the 2 favorites finished 1st or 2nd 81% of the time.

Yesterday I checked how many times the 2 chalks finished in one of the
top 2 top spots & I remember it being 78% of the time.

I THINK YESTERDAY WAS AN A TYPICAL DAY.

I DOWN LOAD THE REAL TIME POOLS AT APPOX 2 MIN TO POST
MOST OF MY GENERATED TOP THREE PICKS ARE THE SAME AS THE PUBLIC'S

THE TOP THREE FINISHED 19 TIMES IN 34 RACES (57% WINNERS), RETURNING A 1.21 ROI FOR AN ODDS ADJUSTED BET.

THAT WAS AN A TYPICAL DAY. TODAY THERE WERE 5 WINS IN 21 RACES.
(24%) FOR A.66 ROI
I POST SOME OF MY PLAYS IN THE "SELECTIONS" FORUM

[ReplayRandall]

CONSIDER THIS.
IF 33% OF THE FAVORITES WIN, THEN 66% OF ALL REMAINING HORSES PLACE.

IF 66% OF THE REMAINING HORSES WIN,
THEN FAVORITE PLACES 33% OF THE TIME.
THEREFOR THE TOP THREE HORSES MUST COMBINE FOR AN EXACTA PAYOUT LESS THAN 32.66% OF THE TIME.

.33*.66*.75 =.1633
.66*.33*.75= .1633
TOTAL .3266
THATS FAVORITE OVER ALL + ALL OVER FAVORITE

What does the takeout have anything to do with the true odds of top3 choices completing the exacta??.....I can't take you seriously anymore, including your foray on trying to beat craps........SEVEN OUT, LINE AWAY....Next shooter, please.

[MONEY]

CONSIDER THIS.
IF 33% OF THE FAVORITES WIN, THEN 66% OF ALL REMAINING HORSES PLACE.

IF 66% OF THE REMAINING HORSES WIN,
THEN FAVORITE PLACES 33% OF THE TIME.
THEREFOR THE TOP THREE HORSES MUST COMBINE FOR AN EXACTA PAYOUT LESS THAN 32.66% OF THE TIME.

.33*.66*.75 =.1633
.66*.33*.75= .1633
TOTAL .3266
THATS FAVORITE OVER ALL + ALL OVER FAVORITE
I don't understand the math, but then I only have a high school education.
Here's how I see it.

If you play the favorite with all and the favorite wins 33% of the time,
you will hit the exacta 33% of the time or 100% of the time that the favorite wins.

If you bet all against the favorite and the favorite loses, but does not finish 2nd
you will lose your money. Today the favorite finished 2nd 19% of the time which, from more
than forty years of losing at the ponies, I know is about average.

So in my book if I bet the favorite over all & all over the favorite I will hit an exacta 51% of the time.
With the median field size being 8 horses the $2. exactas will average $28.00 per race.
So in order to break even the exactas that I hit will have to average $56.00.

Everybody please take into consideration that if you chose to respond to my post, that I am not too bright.

[formula_2002]

What does the takeout have anything to do with the true odds of top3 choices completing the exacta??.....I can't take you seriously anymore, including your foray on trying to beat craps........SEVEN OUT, LINE AWAY....Next shooter, please.

the true odds on a 2-1 horse is not 2-1, you need to adjust for the the takeout.
If the true odds were 2-1 you could make a profit with, say a 1% rebate.

[Jeff P]

Here's what I have in my 2015 database, thoroughbred only, all track codes, Jan 01, 2015 current through yesterday Nov 02, 2015.

First, the sample itself broken out by WPS:

Data Window Settings:
Connected to: C:\JCapper\exe\JCapper2.mdb
999 Divisor Odds Cap: None

SQL: SELECT * FROM STARTERHISTORY
WHERE [DATE] >= #01-01-2015#
AND [DATE] <= #11-02-2015#
ORDER BY [DATE], TRACK, RACE


Data Summary Win Place Show
-----------------------------------------------------
Mutuel Totals 430842.40 431196.20 428768.50
Bet -569828.00 -569828.00 -569828.00
-----------------------------------------------------
P/L -138985.60 -138631.80 -141059.50

Wins 37431 74501 108619
Plays 284914 284914 284914
PCT .1314 .2615 .3812

ROI 0.7561 0.7567 0.7525
Avg Mut 11.51 5.79 3.95

Next, the above sample expressed as exacta matrices for each of the top three ranked by final odds
with no attempt to break ties in the odds, account for dead heats, or account for multiple horses coupled as entries:

Odds Rank = 1 Exacta Matrix
Description Hits Plays Pct Collected Bet P/L Roi
----------------------------------------------------------------------------------
FAV1/FAV2 004446 038381 0.1158 060040.80 -076762.00 -016721.20 0.7822
FAV2/FAV1 003033 038381 0.0790 058331.50 -076762.00 -018430.50 0.7599
FAV1/FAV3 003147 038376 0.0820 059138.50 -076752.00 -017613.50 0.7705
FAV3/FAV1 001865 038376 0.0486 057785.10 -076752.00 -018966.90 0.7529
----------------------------------------------------------------------------------
SUBTOTAL 12,491

Odds Rank = 2 Exacta Matrix
Description Hits Plays Pct Collected Bet P/L Roi
----------------------------------------------------------------------------------
FAV2/FAV1 003109 036697 0.0847 060058.00 -073394.00 -013336.00 0.8183
FAV1/FAV2 004567 036697 0.1245 062141.70 -073394.00 -011252.30 0.8467
FAV2/FAV3 001537 036692 0.0419 053375.60 -073384.00 -020008.40 0.7273
FAV3/FAV2 001301 036692 0.0355 053818.20 -073384.00 -019565.80 0.7334
----------------------------------------------------------------------------------
SUBTOTAL 10,514

Odds Rank = 3 Exacta Matrix
Description Hits Plays Pct Collected Bet P/L Roi
----------------------------------------------------------------------------------
FAV3/FAV1 001843 036588 0.0504 057803.00 -073176.00 -015373.00 0.7899
FAV1/FAV3 003104 036588 0.0848 058503.20 -073176.00 -014672.80 0.7995
FAV3/FAV2 001325 036588 0.0362 055031.60 -073176.00 -018144.40 0.7520
FAV2/FAV3 001571 036588 0.0429 054662.30 -073176.00 -018513.70 0.7470
----------------------------------------------------------------------------------
SUBTOTAL 7,843

----------------------------------------------------------------------------------
TOTAL 30,848 races where two of the top three favorites made the exacta

From a ytd 2015 sample of approx 37,431 races, there were 30,848 races two of the top three favorites made the exacta.

percentage = 30,848 / 37,431

or

percentage = 82.41%


Note - how to read the exacta matrices:

The row designated by the characters FAV1/FAV2 contains data for races where the post time favorite won and the second favorite finished second and exacta wagering was offered. The row designated by the characters FAV1/FAV3 contains data for races where the post time favorite won and the third favorite finished second and exacta wagering was offered, etc.


-jp

.

[formula_2002]

What does the takeout have anything to do with the true odds of top3 choices completing the exacta??.....I can't take you seriously anymore, including your foray on trying to beat craps........SEVEN OUT, LINE AWAY....Next shooter, please.
read my tag line about "luck", it refers to craps where the vig is 1.41%

[AndyC]

TOTAL 30,848 races where two of the top three favorites made the exacta[/CODE]

From a ytd 2015 sample of approx 37,431 races, there were 30,848 races two of the top three favorites made the exacta.

percentage = 30,848 / 37,431

or

percentage = 82.41%

-jp

.

82.41% where 2 of the top 3 favorites made the exacta is IMPOSSBLE!

[whodoyoulike]

CONSIDER THIS.
IF 33% OF THE FAVORITES WIN, THEN 66% OF ALL REMAINING HORSES PLACE.

IF 66% OF THE REMAINING HORSES WIN,
THEN FAVORITE PLACES 33% OF THE TIME.
THEREFOR THE TOP THREE HORSES MUST COMBINE FOR AN EXACTA PAYOUT LESS THAN 32.66% OF THE TIME.

.33*.66*.75 =.1633
.66*.33*.75= .1633
TOTAL .3266
THATS FAVORITE OVER ALL + ALL OVER FAVORITE

Sorry I'm uncertain what you're getting at but, the bolded part doesn't appear correct. I don't know what the % is but it seems important to your formula.

[Jeff P]

82.41% where 2 of the top 3 favorites made the exacta is IMPOSSBLE!
You're right of course.

I see my mistake. I included individual combinations more than once while cutting and pasting rows from the matrices auto generated by my software.

Let's try this again...

Here's the basic sample from my 2015 database, thoroughbred only, all track codes, Jan 01, 2015 current through yesterday Nov 02, 2015:

Data Window Settings:
Connected to: C:\JCapper\exe\JCapper2.mdb
999 Divisor Odds Cap: None

SQL: SELECT * FROM STARTERHISTORY
WHERE [DATE] >= #01-01-2015#
AND [DATE] <= #11-02-2015#
ORDER BY [DATE], TRACK, RACE


Data Summary Win Place Show
-----------------------------------------------------
Mutuel Totals 430842.40 431196.20 428768.50
Bet -569828.00 -569828.00 -569828.00
-----------------------------------------------------
P/L -138985.60 -138631.80 -141059.50

Wins 37431 74501 108619
Plays 284914 284914 284914
PCT .1314 .2615 .3812

ROI 0.7561 0.7567 0.7525
Avg Mut 11.51 5.79 3.95

Next, the above sample expressed as exacta matrices for each of the top three ranked by final odds
with no attempt to break ties in the odds, account for dead heats, or account for coupled entries...

But this time I've included each possible combination among the top three in the odds once only:Odds Rank = 1 Exacta Matrix
Description Hits Plays Pct Collected Bet P/L Roi
----------------------------------------------------------------------------------
FAV1/FAV2 004446 038381 0.1158 060040.80 -076762.00 -016721.20 0.7822
FAV1/FAV3 003147 038376 0.0820 059138.50 -076752.00 -017613.50 0.7705
----------------------------------------------------------------------------------
SUBTOTAL 7,593


Odds Rank = 2 Exacta Matrix
Description Hits Plays Pct Collected Bet P/L Roi
----------------------------------------------------------------------------------
FAV2/FAV1 003109 036697 0.0847 060058.00 -073394.00 -013336.00 0.8183
FAV2/FAV3 001537 036692 0.0419 053375.60 -073384.00 -020008.40 0.7273
----------------------------------------------------------------------------------
SUBTOTAL 4,646


Odds Rank = 3 Exacta Matrix
Description Hits Plays Pct Collected Bet P/L Roi
----------------------------------------------------------------------------------
FAV3/FAV1 001843 036588 0.0504 057803.00 -073176.00 -015373.00 0.7899
FAV3/FAV2 001325 036588 0.0362 055031.60 -073176.00 -018144.40 0.7520
----------------------------------------------------------------------------------
SUBTOTAL 3,168


----------------------------------------------------------------------------------
TOTAL 15,407 races where two of the top three favorites made the exacta


From a ytd 2015 sample of approx 37,431 races, there were 15,407 races where two of the top three favorites made the exacta.

percentage = 15,407 / 37,431

or

percentage = 41.61%


Note - how to read the exacta matrices:

The row designated by the characters FAV1/FAV2 contains data for races where the post time favorite won and the second favorite finished second and exacta wagering was offered. The row designated by the characters FAV1/FAV3 contains data for races where the post time favorite won and the third favorite finished second and exacta wagering was offered, etc.



-jp

.

[lansdale]

Here's what I have in my 2015 database, thoroughbred only, all track codes, Jan 01, 2015 current through yesterday Nov 02, 2015.

First, the sample itself broken out by WPS:

Data Window Settings:
Connected to: C:\JCapper\exe\JCapper2.mdb
999 Divisor Odds Cap: None

SQL: SELECT * FROM STARTERHISTORY
WHERE [DATE] >= #01-01-2015#
AND [DATE] <= #11-02-2015#
ORDER BY [DATE], TRACK, RACE


Data Summary Win Place Show
-----------------------------------------------------
Mutuel Totals 430842.40 431196.20 428768.50
Bet -569828.00 -569828.00 -569828.00
-----------------------------------------------------
P/L -138985.60 -138631.80 -141059.50

Wins 37431 74501 108619
Plays 284914 284914 284914
PCT .1314 .2615 .3812

ROI 0.7561 0.7567 0.7525
Avg Mut 11.51 5.79 3.95

Next, the above sample expressed as exacta matrices for each of the top three ranked by final odds
with no attempt to break ties in the odds, account for dead heats, or account for multiple horses coupled as entries:

Odds Rank = 1 Exacta Matrix
Description Hits Plays Pct Collected Bet P/L Roi
----------------------------------------------------------------------------------
FAV1/FAV2 004446 038381 0.1158 060040.80 -076762.00 -016721.20 0.7822
FAV2/FAV1 003033 038381 0.0790 058331.50 -076762.00 -018430.50 0.7599
FAV1/FAV3 003147 038376 0.0820 059138.50 -076752.00 -017613.50 0.7705
FAV3/FAV1 001865 038376 0.0486 057785.10 -076752.00 -018966.90 0.7529
----------------------------------------------------------------------------------
SUBTOTAL 12,491

Odds Rank = 2 Exacta Matrix
Description Hits Plays Pct Collected Bet P/L Roi
----------------------------------------------------------------------------------
FAV2/FAV1 003109 036697 0.0847 060058.00 -073394.00 -013336.00 0.8183
FAV1/FAV2 004567 036697 0.1245 062141.70 -073394.00 -011252.30 0.8467
FAV2/FAV3 001537 036692 0.0419 053375.60 -073384.00 -020008.40 0.7273
FAV3/FAV2 001301 036692 0.0355 053818.20 -073384.00 -019565.80 0.7334
----------------------------------------------------------------------------------
SUBTOTAL 10,514

Odds Rank = 3 Exacta Matrix
Description Hits Plays Pct Collected Bet P/L Roi
----------------------------------------------------------------------------------
FAV3/FAV1 001843 036588 0.0504 057803.00 -073176.00 -015373.00 0.7899
FAV1/FAV3 003104 036588 0.0848 058503.20 -073176.00 -014672.80 0.7995
FAV3/FAV2 001325 036588 0.0362 055031.60 -073176.00 -018144.40 0.7520
FAV2/FAV3 001571 036588 0.0429 054662.30 -073176.00 -018513.70 0.7470
----------------------------------------------------------------------------------
SUBTOTAL 7,843

----------------------------------------------------------------------------------
TOTAL 30,848 races where two of the top three favorites made the exacta

From a ytd 2015 sample of approx 37,431 races, there were 30,848 races two of the top three favorites made the exacta.

percentage = 30,848 / 37,431

or

percentage = 82.41%


Note - how to read the exacta matrices:

The row designated by the characters FAV1/FAV2 contains data for races where the post time favorite won and the second favorite finished second and exacta wagering was offered. The row designated by the characters FAV1/FAV3 contains data for races where the post time favorite won and the third favorite finished second and exacta wagering was offered, etc.


-jp

.

Hi Jeff,

Much thanks for posting this, but as much an admirer I am of yours, I can't see how this 82% figure could possibly be right. Just by observation, if this were true, the mean mutuel payoff for any exacta combination of the top 3 odds horses would have to be something like $3.20 for $2 - that's for 82% of all exactas. It also seems completely uncorrelated with the win frequencies of the top three horses as I've known them - and similar to what Robert F. posted above. Certainly possible that those frequencies may have increased in the past few years when I've been away from the game, but not this much.

Unless I'm mistaken, the $11.50 average mutuel is for all winning horses, not for the average payoff for a top three exacta box. If you do have this latter figure, it would be very helpful to have, after all this discussion, to maybe reverse engineer the top three exacta winning percentage. However, if you're really right about this percentage, I'd like to figure out how it was arrived out.

Thanks again for your help.

Best,

lansdale

[lansdale]

You're right of course.

I see my mistake. I included individual combinations more than once while cutting and pasting rows from the matrices auto generated by my software.

Let's try this again...

Here's the basic sample from my 2015 database, thoroughbred only, all track codes, Jan 01, 2015 current through yesterday Nov 02, 2015:

Data Window Settings:
Connected to: C:\JCapper\exe\JCapper2.mdb
999 Divisor Odds Cap: None

SQL: SELECT * FROM STARTERHISTORY
WHERE [DATE] >= #01-01-2015#
AND [DATE] <= #11-02-2015#
ORDER BY [DATE], TRACK, RACE


Data Summary Win Place Show
-----------------------------------------------------
Mutuel Totals 430842.40 431196.20 428768.50
Bet -569828.00 -569828.00 -569828.00
-----------------------------------------------------
P/L -138985.60 -138631.80 -141059.50

Wins 37431 74501 108619
Plays 284914 284914 284914
PCT .1314 .2615 .3812

ROI 0.7561 0.7567 0.7525
Avg Mut 11.51 5.79 3.95

Next, the above sample expressed as exacta matrices for each of the top three ranked by final odds
with no attempt to break ties in the odds, account for dead heats, or account for coupled entries...

But this time I've included each possible combination among the top three in the odds once only:Odds Rank = 1 Exacta Matrix
Description Hits Plays Pct Collected Bet P/L Roi
----------------------------------------------------------------------------------
FAV1/FAV2 004446 038381 0.1158 060040.80 -076762.00 -016721.20 0.7822
FAV1/FAV3 003147 038376 0.0820 059138.50 -076752.00 -017613.50 0.7705
----------------------------------------------------------------------------------
SUBTOTAL 7,593


Odds Rank = 2 Exacta Matrix
Description Hits Plays Pct Collected Bet P/L Roi
----------------------------------------------------------------------------------
FAV2/FAV1 003109 036697 0.0847 060058.00 -073394.00 -013336.00 0.8183
FAV2/FAV3 001537 036692 0.0419 053375.60 -073384.00 -020008.40 0.7273
----------------------------------------------------------------------------------
SUBTOTAL 4,646


Odds Rank = 3 Exacta Matrix
Description Hits Plays Pct Collected Bet P/L Roi
----------------------------------------------------------------------------------
FAV3/FAV1 001843 036588 0.0504 057803.00 -073176.00 -015373.00 0.7899
FAV3/FAV2 001325 036588 0.0362 055031.60 -073176.00 -018144.40 0.7520
----------------------------------------------------------------------------------
SUBTOTAL 3,168


----------------------------------------------------------------------------------
TOTAL 15,407 races where two of the top three favorites made the exacta


From a ytd 2015 sample of approx 37,431 races, there were 15,407 races where two of the top three favorites made the exacta.

percentage = 15,407 / 37,431

or

percentage = 41.61%


Note - how to read the exacta matrices:

The row designated by the characters FAV1/FAV2 contains data for races where the post time favorite won and the second favorite finished second and exacta wagering was offered. The row designated by the characters FAV1/FAV3 contains data for races where the post time favorite won and the third favorite finished second and exacta wagering was offered, etc.



-jp

.

Thanks for posting - looks like we were posting at the same time. These frequencies still double and triple my estimates - amazing. They must be increasing big time.

[ReplayRandall]

From a ytd 2015 sample of approx 37,431 races, there were 15,407 races where two of the top three favorites made the exacta.

percentage = 15,407 / 37,431

or

percentage = 41.61%

Thanks Jeff for sharing the work, much appreciated.

Based on over 37K races, we now know what the "exacta average" must pay
to begin profitability. That average price is $14.60 based on a $1 exacta.....
Top3-$1 exacta box= $6 x 100 races= $600

41.61 races won x $14.60= $607.50

Now, the data Jeff used needs to be filtered, eliminating under-laid combos, which reduces the exacta average needed and the hit rate, by whatever means anyone chooses to get to a balanced/averaged payoff rate, while not depressing the strike rate to overall ROI unprofitability.

Of course, this will be back-fitting, but we must have a start point of reference to begin, then going forward with another 5000 races to see if severe odds flux as horses load, causes the whole methodology to become fruitless from the start, as pointed out by previous posters.....

[RXB]

About 42% for the two of the top three wagering choices sweeping the exacta.

About 49%, one of the top three combine with a fourth or lower choice to fill the exacta.

About 9%, the fourth and lower choices sweep the exacta.

But these probabilities (and the resulting payouts) are so hugely dependent on field size, an overall percentage and average payout for the top three wagering choices doesn't really mean much.

[appistappis]

read my tag line about "luck", it refers to craps where the vig is 1.41%

I'm sure everyone here knows one or two people who play this game for a living.....I have been a craps dealer for 41 years and have never seen a long term winner.

[PICSIX]

I'm sure everyone here knows one or two people who play this game for a living.....I have been a craps dealer for 41 years and have never seen a long term winner.

Just curious, what is the most times you've seen :7: rolled in a row in your career?

Thanks,

Mike

[DeltaLover]

You're right of course.

I see my mistake. I included individual combinations more than once while cutting and pasting rows from the matrices auto generated by my software.

Let's try this again...

Here's the basic sample from my 2015 database, thoroughbred only, all track codes, Jan 01, 2015 current through yesterday Nov 02, 2015:

Data Window Settings:
Connected to: C:\JCapper\exe\JCapper2.mdb
999 Divisor Odds Cap: None

SQL: SELECT * FROM STARTERHISTORY
WHERE [DATE] >= #01-01-2015#
AND [DATE] <= #11-02-2015#
ORDER BY [DATE], TRACK, RACE


Data Summary Win Place Show
-----------------------------------------------------
Mutuel Totals 430842.40 431196.20 428768.50
Bet -569828.00 -569828.00 -569828.00
-----------------------------------------------------
P/L -138985.60 -138631.80 -141059.50

Wins 37431 74501 108619
Plays 284914 284914 284914
PCT .1314 .2615 .3812

ROI 0.7561 0.7567 0.7525
Avg Mut 11.51 5.79 3.95

Next, the above sample expressed as exacta matrices for each of the top three ranked by final odds
with no attempt to break ties in the odds, account for dead heats, or account for coupled entries...

But this time I've included each possible combination among the top three in the odds once only:Odds Rank = 1 Exacta Matrix
Description Hits Plays Pct Collected Bet P/L Roi
----------------------------------------------------------------------------------
FAV1/FAV2 004446 038381 0.1158 060040.80 -076762.00 -016721.20 0.7822
FAV1/FAV3 003147 038376 0.0820 059138.50 -076752.00 -017613.50 0.7705
----------------------------------------------------------------------------------
SUBTOTAL 7,593


Odds Rank = 2 Exacta Matrix
Description Hits Plays Pct Collected Bet P/L Roi
----------------------------------------------------------------------------------
FAV2/FAV1 003109 036697 0.0847 060058.00 -073394.00 -013336.00 0.8183
FAV2/FAV3 001537 036692 0.0419 053375.60 -073384.00 -020008.40 0.7273
----------------------------------------------------------------------------------
SUBTOTAL 4,646


Odds Rank = 3 Exacta Matrix
Description Hits Plays Pct Collected Bet P/L Roi
----------------------------------------------------------------------------------
FAV3/FAV1 001843 036588 0.0504 057803.00 -073176.00 -015373.00 0.7899
FAV3/FAV2 001325 036588 0.0362 055031.60 -073176.00 -018144.40 0.7520
----------------------------------------------------------------------------------
SUBTOTAL 3,168


----------------------------------------------------------------------------------
TOTAL 15,407 races where two of the top three favorites made the exacta


From a ytd 2015 sample of approx 37,431 races, there were 15,407 races where two of the top three favorites made the exacta.

percentage = 15,407 / 37,431

or

percentage = 41.61%


Note - how to read the exacta matrices:

The row designated by the characters FAV1/FAV2 contains data for races where the post time favorite won and the second favorite finished second and exacta wagering was offered. The row designated by the characters FAV1/FAV3 contains data for races where the post time favorite won and the third favorite finished second and exacta wagering was offered, etc.



-jp

.

:ThmbUp: :ThmbUp:

Very accurate posting (as a side note, I think that we need more of these rather than having 200 pages discussing the conspiracies behind AP BCC win and all the possible wouldas couldas shouldas)

I have to admit that the 41% looked a bit high to me but I was able to duplicate it.

For the history the related SQL code I wrote to collect the data is the following:


set @num := 1, @race_id := 0, @all_races=0;

select @all_races := count(*) from
(
select
distinct(race_id)
from
race_starters
where
program_number != 'SCR' and
race_id > 0
) as all_races_tmp_table

select
@all_races as 'Races',
count(*) as 'Matches',
count(*) * 100 / @all_races as 'Percentage'
from
(select race_id, count(*) as counter from
(select
race_id,
finish_position,
ordered_by_odds
from (
select
race_id,
odds,
finish_position,
@num := if(@race_id = race_id, @num + 1, 1) as ordered_by_odds,
@race_id := race_id as dummy
from
race_starters
where
program_number != 'SCR' and
race_id > 0
order by
race_id,
odds
) as temp
where
finish_position in (1,2) and
ordered_by_odds <=3
order by race_id) as temp3
group by race_id
having counter = 2) as temp4


while the results look like this:

http://i67.tinypic.com/vhgklw.png

[Robert Fischer]

using the data that Jeff P was so kind to share with us, I added the unique combinations and the result was 40%.

So, our guesses were not that far off.

here are the values added to reach 40%:
combination hit%
1/2 0.1158
2/1 0.0790
1/3 0.0820
3/1 0.0486
2/3 0.0419
3/2 0.0355

.4028

40%



(EDIT: looks like Delta, RXB and perhaps others have also presented similar estimates :ThmbUp: )

[Actor]

over a large number of races, say 500, how often do two of the top three horses come in 1st and 2nd ?



appreciate your answer.45% of the time using a database of 1496 races.

[davew]

CONSIDER THIS.
IF 33% OF THE FAVORITES WIN, THEN 66% OF ALL REMAINING HORSES PLACE.

IF 66% OF THE REMAINING HORSES WIN,
THEN FAVORITE PLACES 33% OF THE TIME.
THEREFOR THE TOP THREE HORSES MUST COMBINE FOR AN EXACTA PAYOUT LESS THAN 32.66% OF THE TIME.

.33*.66*.75 =.1633
.66*.33*.75= .1633
TOTAL .3266
THATS FAVORITE OVER ALL + ALL OVER FAVORITE


you forgot the 2/3 and 3/2 combinations which also are included in your original question.

[formula_2002]

you forgot the 2/3 and 3/2 combinations which also are included in your original question.


Yes, I woke up in the middle of the night and realized the omission. :)
Not sure how to handle that 2/3 and 3/2

Having said that, the 1/2 and 2/1 needs additional analysis because the 1/2 in more weighted than the 2/1.

Regardless, Jeff P had everything I was asking for.

[formula_2002]

Jeff P. thanks for the valued information. HOOYA!!

Joe M

[lansdale]

using the data that Jeff P was so kind to share with us, I added the unique combinations and the result was 40%.

So, our guesses were not that far off.

here are the values added to reach 40%:
combination hit%
1/2 0.1158
2/1 0.0790
1/3 0.0820
3/1 0.0486
2/3 0.0419
3/2 0.0355

.4028

40%



(EDIT: looks like Delta, RXB and perhaps others have also presented similar estimates :ThmbUp: )

Much thanks for breaking the combinations out. As we can see ca. 28% of the handicapping weight is in the top three combinations. What is mind-blowing to me is the increased hit-rate of these ranks since I left close contact with the game ca. 2010. Just guessing, it appears that the biggest changes are that first and second favorites place frequencies have increased to ca. 33% - huge jumps from what I posted. Would love to see the actual frequencies.

The WPS frequencies I posted above were roughly accurate for ca. 2007-2010, and your figures appear somewhat similar. I think these were historically accurate. If you take a look at this 2003 post from Barry Meadow, in which he provides WPS figures for favorites, they are also consistent with those figures. So, we can how drastically the numbers have changed since that time. A quantum leap.

Old 06-27-2008, 03:36 AM #2
BMeadow
Registered User

Join Date: May 2001
Posts: 191
vCash: 400
We ran several surveys in Meadow's Racing Monthly regarding these questions. The surveys were compiled by Jim Cramer of Handicapper's Data Warehouse.

A survey of 200,000 post-time favorites that appeared in the April 2003 issue showed the public favorite winning 33.6% with an ROI of 0.82. A previous survey of a different 200,000 post-time favorites showed 32.8% wins, 20.4% place, and 14.0% show with an ROI of 0.81.

We also did a morning-line survey of 46 tracks with at least 2,000 morning line favorites. At every single track, post-time favorites won more often than morning-line favorites. Overall, morning-line favorites won a bit less than 30% with an ROI of 0.81.

I have never seen any way to use the morning line to win; it is simply one employee's prediction of how the public will bet. Horses who go off at post time at 2-1 win approximately 28% no matter whether they are 4-5 in the morning line or 15-1.
__________________
Barry Meadow
BMeadow is offline Report Bad Post

Assuming these exacta figures which have been corroborated buy a few posters her, are current reality, they imply that the favorite is ITM in close to 100% of races, and the second favorite in 75%. Likely, exotics players are already down with this, but it is astonishing to me.

[formula_2002]

over a large number of races, say 500, how often do two of the top three horses come in 1st and 2nd ?



appreciate your answer.

the reason for the question was to find out how my results compared to a large data base

as of this time:
206 races, selected horses for an exacta play (generally top 3,4 or 5)
won 60% of the time for a .92 roi.

71 races ago I decided to break out the top 3.
46.5% winners returned a 1.14 roi

[raybo]

Today's tracks

Finger Lakes
Hawthorne
Mahoning Valley
Parx Racing
Portland Meadows
Turf Paradise
Zia Park

Today there were 63 thoroughbred races.

The off time favorite won 30 times/placed 12 times.

The 2nd favorite won 11 times/placed 12 times.

The 2 favorites combined to make the exacta 13 times.

Neither of the favorites finished 1st or 2nd 11 times.

Today the 2 favorites finished 1st or 2nd 81% of the time.

Yesterday I checked how many times the 2 chalks finished in one of the
top 2 top spots & I remember it being 78% of the time.

Just in that small sample the OP's question has an answer of:

13/63 = .206 (20.6% of the time) Unless I am missing something. There were 63 races, 13 of those resulted in the 1st and 2nd favorites making up the exacta.

[raybo]

Yes...but you have not addressed this thread's original question. :)

Yeah, and I based my 20.6% on his sample. :bang: :lol:

[MONEY]

Just in that small sample the OP's question has an answer of:

13/63 = .206 (20.6% of the time) Unless I am missing something. There were 63 races, 13 of those resulted in the 1st and 2nd favorites making up the exacta.

That is what happened.

[raybo]

Hi Jeff,

Much thanks for posting this, but as much an admirer I am of yours, I can't see how this 82% figure could possibly be right. Just by observation, if this were true, the mean mutuel payoff for any exacta combination of the top 3 odds horses would have to be something like $3.20 for $2 - that's for 82% of all exactas. It also seems completely uncorrelated with the win frequencies of the top three horses as I've known them - and similar to what Robert F. posted above. Certainly possible that those frequencies may have increased in the past few years when I've been away from the game, but not this much.

Unless I'm mistaken, the $11.50 average mutuel is for all winning horses, not for the average payoff for a top three exacta box. If you do have this latter figure, it would be very helpful to have, after all this discussion, to maybe reverse engineer the top three exacta winning percentage. However, if you're really right about this percentage, I'd like to figure out how it was arrived out.

Thanks again for your help.

Best,

lansdale

He already stated that that was incorrect, as he listed some combinations more than once. I saw that when I looked at his DB printout but didn't want to make a fool of myself by challenging Jeff!

[raybo]

That is what happened.

Yes, but the OP's question was what percentage of the time do 2 of the top 3 favorites make up the exacta, not just the top 2 favorites.

[JJMartin]

Don't forget you cannot accurately pick the top 3 positions 100% of the time because they are not always known until after the race is over. The 3rd fav for example could slip into 4th after your ticket is bet. So if you tried to bet a system based on this data, you would have frequent failure.

[raybo]

My personal opinion is that if you base any system on the top 3 or 4 favorites, you cannot box them all. At some point you have to make a stand on a single horse on top or bottom, or you have to wait until the favorite can be taken off the ticket completely. I've run these scenarios through tests many times in the distant past and the costs, if boxed, are just too high to make long term profit.

[formula_2002]

My personal opinion is that if you base any system on the top 3 or 4 favorites, you cannot box them all. At some point you have to make a stand on a single horse on top or bottom, or you have to wait until the favorite can be taken off the ticket completely. I've run these scenarios through tests many times in the distant past and the costs, if boxed, are just too high to make long term profit.

Raybo, the idea is not to box, rather use a proportional odds bet based on some formula that predicts a bet size which beats the takeout and makes a profit.
I use the live tote board data which I down load at 2 minuets to post time.

Of course I've been doing this for about 10 years. Previous to that I used a hand calculator.

Some day I hope to get this thing profitable. :)

Until then it keeps me from make bad bets..

[formula_2002]

Jeff, how come the number of plays for each combination does not equal the number of races, 37431 ?


based on 37431 races
combination hits plays percent collected bet roi
fav1/fav2 4446 38382 0.12 60040 76762 0.78
fav1/fav3 3147 38376 0.08 59138 76752 0.77
fav2/fav1 3109 36697 0.08 60058 73394 0.82
fav2/fav3 1537 36692 0.04 53375 73384 0.73
fav3/fav1 1843 36588 0.05 57803 73176 0.79
fav3/fav2 1325 36588 0.04 55031 73176 0.75
41.10% 345445 446644
% win 0.77 roi

[raybo]

Jeff, how come the number of plays for each combination does not equal the number of races, 37431 ?


based on 37431 races
combination hits plays percent collected bet roi
fav1/fav2 4446 38382 0.12 60040 76762 0.78
fav1/fav3 3147 38376 0.08 59138 76752 0.77
fav2/fav1 3109 36697 0.08 60058 73394 0.82
fav2/fav3 1537 36692 0.04 53375 73384 0.73
fav3/fav1 1843 36588 0.05 57803 73176 0.79
fav3/fav2 1325 36588 0.04 55031 73176 0.75
41.10% 345445 446644
% win 0.77 roi


Wouldn't dead heats explain the fav1/fav2 and fav1/fav3 larger number of plays than races? Maybe the others with smaller numbers of plays than races would be races that didn't offer exactas, match races/extremely small fields, etc.?

[thespaah]

over a large number of races, say 500, how often do two of the top three horses come in 1st and 2nd ?



appreciate your answer.
I'll take a shot....Since winning favs hit the top spot about in about 1/3 rd of all races and ITM favs run at about a 65% clip, one would think this happens at least twice or three times on every card,....Well
I checked.. I chose at random two days Bel, Med and Lrl..
that's six race cards....
Over the 40 or so races (Med has 6 races on the program) it happened.....TWICE...
Once at Bel, once at Med..
That's it.
Now, at tracks with typically smaller fields and our number one track for an endless parade of short priced horses, Finger Lakes, I suppose this occur a bit more often...

[thespaah]

top three horses by what criteria? how many horses in race? class level? surface? ....
I just went by post time odds. That's what I thought the OP was asking.
Because the term "best horse" is an opinion of the individual handicapping the races....

[DeltaLover]

Raybo, the idea is not to box, rather use a proportional odds bet based on some formula that predicts a bet size which beats the takeout and makes a profit.
I use the live tote board data which I down load at 2 minuets to post time.

Of course I've been doing this for about 10 years. Previous to that I used a hand calculator.

Some day I hope to get this thing profitable. :)

Until then it keeps me from make bad bets..

For more than one reasons what you describe here is impossible to accomplish. Some of them are as follows:

(1) It is impossible to know the final order of horses based on their odds

(2) Mechanically betting the same combos based on the odds ranking is impossible to beat the rake which will destroy even the largest bankroll very quickly

(3) The offered payout for each exacta combo in never known before the closing of the race

[green80]

[QUOTE=formula_2002] over a large number of races, say 500, how often do two of the top three horses come in 1st and 2nd ?




Not often enough to make any money boxing them.

[formula_2002]

For more than one reasons what you describe here is impossible to accomplish. Some of them are as follows:

(1) It is impossible to know the final order of horses based on their odds

(2) Mechanically betting the same combos based on the odds ranking is impossible to beat the rake which will destroy even the largest bankroll very quickly

(3) The offered payout for each exacta combo in never known before the closing of the race

my system results so far. it's based on win and exacta pool data downloaded 2 min. to post. my ranking can sometimes differ from the final odds ranking. these are not boxed bets. the bet size is proportional to the my calculated odds.


RESULTS
AN EXACTA BET ON ALL CALCULATED TOP RANKED COMBINATIONS (SOME TIMES THERE ARE BUT 2 PICKS IN A RACE, SOME TIMES 6, MOSTLY 4 TO 5)


275 RACES
.92 ROI
59% WINS

TOP 4
108
.94 ROI
55% WINS

TOP 3
126 RACES
.97 ROI
43% WINS
STARTED TRACKING TOP 3 AFTER RACE #136

[formula_2002]

I'll take a shot....Since winning favs hit the top spot about in about 1/3 rd of all races and ITM favs run at about a 65% clip, one would think this happens at least twice or three times on every card,....Well
I checked.. I chose at random two days Bel, Med and Lrl..
that's six race cards....
Over the 40 or so races (Med has 6 races on the program) it happened.....TWICE...
Once at Bel, once at Med..
That's it.
Now, at tracks with typically smaller fields and our number one track for an endless parade of short priced horses, Finger Lakes, I suppose this occur a bit more often...


After looking at it again, the % winner changes less than 1/2 percent.
That's the advantage working with a large data base.