Beyer figs

[sbcaris]

Beyer figs have been bashed by many but they still are a true indicator of how a horse runs. In fact the impact value for using the highest or second highest Beyer figure achieved in a horses last race before the Derby at 9 furlongs on dirt is well above average.

I did a study of the last 12 years of Ky Derbies and recorded the highest Beyer fig in each year's Derby fields for a horse's last race at 9 furlongs and the second highest Beyer fig in his last race at 9 furlongs for each year. Here are the results of that study:

2014--Highest-California Chrome (WON)
Second Highest-Out of the money

2013--Highest-Goldencents -Out of the Money
Second Highest Orb-WON

2012--Highest-Bodemeister (Second)
Second Highest ---Out of the Money

2011- Highest -Nehro--(Second)
Second Highest was out of the money

2010- Highest -Ice Box (Second)
Second Highest Super Saver (WON)

2009 Highest and Second Highest were Out of the Money

2008-Highest Big Brown WON
Second Highest was out of the money

2007- Highest Curlin (Third)
Second Highest was out of the money

2006-Highest and Second Highest were out of the money

2005- Highest was out of the money
Second Highest was Afleet Alex who ran third

2004-Highest and second highest were out of the money

2003 Highest Empire Maker ran (Second)
Second Highest Funny Cide (WON)

A statistical analysis of the above data follows: 5 winners in 12 years equals a win percentage of 41.6%. Since the percent of starters is 10.4% the impact value to win the roses is a strong 4.00 which means horses who achieve the highest or second highest Beyer at 9 furlongs in their last start before Derby day are winning the roses 4 times more often than statistical expectation.

Since 4 of the 12 ran second the percentage of second place finishers is 33.3%. Since the percent of starters qualifying on this angle is 10.4 the impact value for finishing second is also strong at 3.20. Horses with the highest or second highest Beyer fig in their respective Derby fields are running second in the roses more than 3 times statistical expectation.

So much for bashing the Beyer figs in the Derby. The horses with the best or second best are doing very well.

[Lemon Drop Husker]

I used to have a firm rule that a horse had to achieve at least a 100 Beyer at a mile or more distance in order for me to consider a Win bet in the Derby.

I stopped with that "firm rule" about 7 or 8 years ago as it became apparent that in this new Derby age that figure was no longer useful for handicapping purposes.

I do believe that Beyer numbers have to be considered, but they don't hold any firm rules or high end percentages like they used to hold.

[SecretAgentMan]

I used to have a firm rule that a horse had to achieve at least a 100 Beyer at a mile or more distance in order for me to consider a Win bet in the Derby.

I stopped with that "firm rule" about 7 or 8 years ago as it became apparent that in this new Derby age that figure was no longer useful for handicapping purposes.

I do believe that Beyer numbers have to be considered, but they don't hold any firm rules or high end percentages like they used to hold.


^^^^

This.......I agree with Husker, I always look at Bayer figures but don't make them a must to play a horse. They use to be more of a sure way to pick a derby winner in the 90's & early 2,000's.......

I definitely still look at what beyer figure horses have received. What crazy is the high beyer the horses were receiving last year at GP because the track was sizzling.......this year its like molasses & Upstart & Materiality received big beyers as well.

[sbcaris]

I do not make anything a must. There are no hard and fast rules that always work. However, the data I supplied above indicates that horses who run the highest or second highest Beyer figs at 9 furlongs on dirt in their last start before the Derby have done very well by finishing first or second much better than statistical expectation.

It is very difficult to set hard and fast rules because the Derby landscape keeps changing. In Super Saver's year every horse ran less than a 100 in his final 9 furlong prep. Naturally the winner will come in under 100 as well. The highest Beyer that year was Ice Box with a 99 and he ran second. The second highest Beyer was Super Saver with a 98 and he WON.

[f2tornado]

Sometimes I wonder if Beyer blows the track variants resulting in an inflated/deflated number.

Santa Anita Derby:
California Chrome - 1:47.52 BSF 107
I'll Have Another - 1:47.88 BSF 95

Arkansas Derby:
Super Saver - 1:49.1 BSF 98
Smarty Jones - 1:49.41 BSF 109 (muddy; but BSF on par w/previous start fast track)

Certainly use the BSFs and BRIS numbers but don't forget about the final times. Those are the most objective figures available. A minute is one minute. I suspect Beyer has blown up Gulfstream figures this year and I might be able to get partial confirmation one way or another this weekend as a few Gulf stakes horses are running at Aqueduct and at least one at Keeneland. Beyer has also been criticized for his figures on poly. I tend to favor the Brisnet figures on that type of track.

If the Santa Anita Derby winner has a final time of 1.47.9 or less then that horse is a strong Derby (and Preakness/Belmont) contender. According to my back of the envelope numbers there have been 9 SA Derby winners with sub 1:47.9 final times since 1980. Five of these have gone on to win the Kentucky Derby. That's a big impact value. Of the four that missed, two hit the show pool, one got bottom of super, and one just missed the modern day exotics coming in 5th. The year Indian Charlie won in 1:47 Real Quiet was just 2 1/2 back (approx 0.5s) so that horse would have also qualified bringing your Derby winner total to six. Free House headed Silver Charm in 1:47.6 so that one qualified too. That's seven winners. I have yet to find a Derby angle with a reasonable rate of recurrence stronger than this one.

[Lemon Drop Husker]

I do not make anything a must. There are no hard and fast rules that always work. However, the data I supplied above indicates that horses who run the highest or second highest Beyer figs at 9 furlongs on dirt in their last start before the Derby have done very well by finishing first or second much better than statistical expectation.

It is very difficult to set hard and fast rules because the Derby landscape keeps changing. In Super Saver's year every horse ran less than a 100 in his final 9 furlong prep. Naturally the winner will come in under 100 as well. The highest Beyer that year was Ice Box with a 99 and he ran second. The second highest Beyer was Super Saver with a 98 and he WON.

Oh, I agree that the Derby landscape is vastly changing. Almost every hard set rule for decades has been broken this millennium with about the only one left being the "Apollo curse".

Giacomo and Mine That Bird broke about 38 "Derby Rules" in and of themselves. :lol:

[letswastemoney]

I only use TimeformUS and haven't touched Beyers for 2 years.

With that said, speed figures are just one tool in the overall picture. I think they are underrated in the Derby though because everyone is focused on pedigree to a larger degree than any other race.

[ReplayRandall]

I only use TimeformUS and haven't touched Beyers for 2 years.

With that said, speed figures are just one tool in the overall picture. I think they are underrated in the Derby though because everyone is focused on pedigree to a larger degree than any other race.
When you first started using TFUS, did you compare the figs between the two for winners, for value, or both? Did you come to the conclusion that TFUS figs were superior or was it a combination of other factors why you abandoned Beyers? BTW, I've used both, among others.....

[sbcaris]

Toronado: I looked over the last 14 years and in that time frame ONLY 3 horses broke 1:48 in the Santa Anita Derby: Point Given who lost the roses in 2001, and Ill Have Another who won the roses in 2012 and California Chrome who won the roses last year.

Breaking 1:47 is a difficult task to say the least. The only 3 runners to do it in the last 14 years won two legs of the Triple Crown. Interesting angle but will Dortmund or Bolo or Prospect Park run that fast Saturday?

[letswastemoney]

When you first started using TFUS, did you compare the figs between the two for winners, for value, or both? Did you come to the conclusion that TFUS figs were superior or was it a combination of other factors why you abandoned Beyers? BTW, I've used both, among others.....I just slowly lost faith in Beyers because I don't always know if they adjust the figure, and I want to know when it's adjusted for pace or other reasons so I could look at the race more closely.

And they don't have pace figures for their normal PPs so I can't see how the pace affected the number on my own.

I don't have any hard statistics, but I believe I'm at least equal or better with TimeformUS. I'm certainly not worse and they're free on TVG anyway, so switching was a no-brainer.

[f2tornado]

Breaking 1:48 (corrected) is a difficult task to say the least. The only 3 runners to do it in the last 14 years won two legs of the Triple Crown. Interesting angle but will Dortmund or Bolo or Prospect Park run that fast Saturday?

Possible if the final half furlong resembles final half of the San Felipe (6.14s). Certainly some speed in that race to set up a brisk pace should they come home a tad slower. Another way to look at the angle is there has been no SA Derby winner win the Roses without a sub 1:48 time since Affirmed hit 1:48 flat back in 1978. In my brief looking, I only found Ferdinand who finished 3rd in slower time won in 1986. Perhaps a method of dumping a favorite off the top of your ticket in years with when final race times are not so fast.

[MPRanger]

I think stakes grade 3yo's can always improve over a previous figure. The SF shows what he can do. Not what he can't do.

[Investorater]

Throw enough studies out there and something is bound to stick. A person will be correct with at least one theory on the KY Derby.
As we all are from time to time. It is all interesting.

[sbcaris]

It's not what works from time to time or only once in a while. It's what works consistently to get winners that matters most and that is why I do plenty of research to come up with angles that have proven to work well in the past and have high impact values and high ROIs. Those kind of angles do not pick the Derby winner every year but over the long haul prove to be golden nuggets for the handicapper. Examples of angles that prove successful in the long haul: final fraction indicators, conduit mares in tail female etc.

[Lemon Drop Husker]

It's not what works from time to time or only once in a while. It's what works consistently to get winners that matters most and that is why I do plenty of research to come up with angles that have proven to work well in the past and have high impact values and high ROIs. Those kind of angles do not pick the Derby winner every year but over the long haul prove to be golden nuggets for the handicapper. Examples of angles that prove successful in the long haul: final fraction indicators, conduit mares in tail female etc.

Yep.

And I love the info you bring. Without any doubt, the Derby does have tendencies.

[SecretAgentMan]

Beyer figs have been bashed by many but they still are a true indicator of how a horse runs. In fact the impact value for using the highest or second highest Beyer figure achieved in a horses last race before the Derby at 9 furlongs on dirt is well above average.

I did a study of the last 12 years of Ky Derbies and recorded the highest Beyer fig in each year's Derby fields for a horse's last race at 9 furlongs and the second highest Beyer fig in his last race at 9 furlongs for each year. Here are the results of that study:

2014--Highest-California Chrome (WON)
Second Highest-Out of the money

2013--Highest-Goldencents -Out of the Money
Second Highest Orb-WON

2012--Highest-Bodemeister (Second)
Second Highest ---Out of the Money

2011- Highest -Nehro--(Second)
Second Highest was out of the money

2010- Highest -Ice Box (Second)
Second Highest Super Saver (WON)

2009 Highest and Second Highest were Out of the Money

2008-Highest Big Brown WON
Second Highest was out of the money

2007- Highest Curlin (Third)
Second Highest was out of the money

2006-Highest and Second Highest were out of the money

2005- Highest was out of the money
Second Highest was Afleet Alex who ran third

2004-Highest and second highest were out of the money

2003 Highest Empire Maker ran (Second)
Second Highest Funny Cide (WON)

A statistical analysis of the above data follows: 5 winners in 12 years equals a win percentage of 41.6%. Since the percent of starters is 10.4% the impact value to win the roses is a strong 4.00 which means horses who achieve the highest or second highest Beyer at 9 furlongs in their last start before Derby day are winning the roses 4 times more often than statistical expectation.

Since 4 of the 12 ran second the percentage of second place finishers is 33.3%. Since the percent of starters qualifying on this angle is 10.4 the impact value for finishing second is also strong at 3.20. Horses with the highest or second highest Beyer fig in their respective Derby fields are running second in the roses more than 3 times statistical expectation.

So much for bashing the Beyer figs in the Derby. The horses with the best or second best are doing very well.


So going by this, as of today, Metriality or Upstart have a shot at winning the KD since they have the highest & 2nd highest beyers.

Of course we still have 4 more big preps to run........if Upstart does win the KD, he can regress & run a 103 or 104 beyer & still win.

[sbcaris]

Yes, Materiality and Upstart have a shot at winning the roses if they get into the starting gate. It looks like they will have the highest and second highest Beyer figs in the field at this point unless American Pharaoh or some other runs a huge number this Saturday in the Arkansas Derby.

Once again, even though the stats show us a strong angle, that angle of first and second best Beyer gets 40% winners which means it also gets 60% losers in the run for the roses. Its a good angle but like all good angles it has drawbacks --it only picks the winner 4 of every 10 Derbies and loses 6 of every 10.

[BlueChip@DRF]

Yes, Materiality and Upstart have a shot at winning the roses if they get into the starting gate. It looks like they will have the highest and second highest Beyer figs in the field at this point unless American Pharaoh or some other runs a huge number this Saturday in the Arkansas Derby.

Once again, even though the stats show us a strong angle, that angle of first and second best Beyer gets 40% winners which means it also gets 60% losers in the run for the roses. Its a good angle but like all good angles it has drawbacks --it only picks the winner 4 of every 10 Derbies and loses 6 of every 10.


I would take a 40% chance to win with 20-1 odds.

[f2tornado]

I would take a 40% chance to win with 20-1 odds.

The sample size is statistically small meaning there is not gonna be a high degree of confidence that 40% figure is representative of the condition. Materiality is 0% using a different metric. Upstart not a bad play at 20-1 if one can look past the snail times in the FOY/FL Derby and I think he'll be close to those odds at post.

[ReplayRandall]

Upstart not a bad play at 20-1 if one can look past the snail times in the FOY/FL Derby and I think he'll be close to those odds at post.
Upstart will be between 10-12 to 1. No way in the world this horse goes off at anything close to 20-1. However, I'm rooting for you to be right on those over-generous odds, as I'll be shoving it in with both hands.. ;)

[SecretAgentMan]

Upstart will be between 10-12 to 1. No way in the world this horse goes off at anything close to 20-1. However, I'm rooting for you to be right on those over-generous odds, as I'll be shoving it in with both hands.. ;)



The fav will be 5-1.......then the other supposed great horses will be between 6-1 to 12-1.......Upstart will be between 15-1 to 20-1

[ReplayRandall]

The fav will be 5-1.......then the other supposed great horses will be between 6-1 to 12-1.......Upstart will be between 15-1 to 20-1
I hope you're right........but sadly you won't be, as too many bettors see the same qualities in Upstart as I do, and will not let him drift that high.

[f2tornado]

While my approximately 20-1 is probably on the high end, never underestimate the SoCal betting crowd to crush a horse like Dortmund and lift the likes of Upstart. Too bad Prospect Park probably doesn't get in as he'd take a pile of money too. El Kabeir could have helped the cause with some New York money but he floundered in the Wood. If Pharoah romps in the Ark as many expect then you're gonna see several solid contenders sitting in the 15 or 20-1 range along with a couple live longshots/exotics bombers in the 30 to 40-1 range.

[BlueChip@DRF]

While my approximately 20-1 is probably on the high end, never underestimate the SoCal betting crowd to crush a horse like Dortmund and lift the likes of Upstart. Too bad Prospect Park probably doesn't get in as he'd take a pile of money too. El Kabeir could have helped the cause with some New York money but he floundered in the Wood. If Pharoah romps in the Ark as many expect then you're gonna see several solid contenders sitting in the 15 or 20-1 range along with a couple live longshots/exotics bombers in the 30 to 40-1 range.


This was an exotics bomb I was hoping for.

[Grits]

Maybe its just me..but I'd hesitate to make the statement that "too bad Prospect Park doesn't get in" when we're still just shy of 4 weeks from post time. Is he not #21 on the points list? If so, its possible he'll be in the gate. Anything can (and does) happen this far out.

[f2tornado]

Maybe its just me..but I'd hesitate to make the statement that "too bad Prospect Park doesn't get in" when we're still just shy of 4 weeks from post time. Is he not #21 on the points list? If so, its possible he'll be in the gate. Anything can (and does) happen this far out.

You misquoted my post. I stated "...probably doesn't get in". I agree there could be a defection or two and perhaps an unfortunate injury but points from the Ark Derby could bump PP down to 24 on the list. I suppose Pharoah could fall over which which could put PP as low as 25. Then, even if the horse has enough points the connections will need to decide if he would run the grueling race after his 4th place finish in the SAD. I hope he gets in as that horse will take some money. I don't think it has a chance but my old man said many times, "Thinking so doesn't make it so".

[sbcaris]

I did a study of all the years in which Beyer figs were published in the Daily Racing Form (1992-2014). Here are the results of that study which includes the horses with the highest last out Beyer fig and the second highest last out Beyer fig:

There were 10 winners in that period: California Chrome, Orb, Super Saver, Big Brown, Funny Cide, War Emblem, Fusaichi Pegasus, Charismatic, Silver Charm, and Lil E Tee.

10 winners in 23 years is 43.4% winners. Since the percent of starters is approximately 14% the impact value for this angle is a strong 3.10. These types with the highest or second highest last out Beyer fig are winning the roses around 3 times more often than statistical expectation. Now I know the sample size is only 23 years but that's all we have to go on because Beyer figs were first published in the Form in 1992.

However, from the above study one can certainly conclude that having the best Beyer fig or the second best Beyer fig in the last race is a positive handicapping angle.

Since there were 7 second place finishers that qualified the impact value for place is also positive at 2.14 (30% second placers divided by 14% of the starters.

[Grits]

You misquoted my post. I stated "...probably doesn't get in". I agree there could be a defection or two and perhaps an unfortunate injury but points from the Ark Derby could bump PP down to 24 on the list. I suppose Pharoah could fall over which which could put PP as low as 25. Then, even if the horse has enough points the connections will need to decide if he would run the grueling race after his 4th place finish in the SAD. I hope he gets in as that horse will take some money. I don't think it has a chance but my old man said many times, "Thinking so doesn't make it so".

I DID misquote you. I'm sorry.

[OCF]

I did a study of all the years in which Beyer figs were published in the Daily Racing Form (1992-2014). Here are the results of that study which includes the horses with the highest last out Beyer fig and the second highest last out Beyer fig:

There were 10 winners in that period: California Chrome, Orb, Super Saver, Big Brown, Funny Cide, War Emblem, Fusaichi Pegasus, Charismatic, Silver Charm, and Lil E Tee.

10 winners in 23 years is 43.4% winners. Since the percent of starters is approximately 14% the impact value for this angle is a strong 3.10. These types with the highest or second highest last out Beyer fig are winning the roses around 3 times more often than statistical expectation. Now I know the sample size is only 23 years but that's all we have to go on because Beyer figs were first published in the Form in 1992.

However, from the above study one can certainly conclude that having the best Beyer fig or the second best Beyer fig in the last race is a positive handicapping angle.

Since there were 7 second place finishers that qualified the impact value for place is also positive at 2.14 (30% second placers divided by 14% of the starters.

Great stuff, proposed angles are much more meaningful when accompanied by impact values (strong ones, of course!).

[rastajenk]

There were 10 winners in that period: California Chrome, Orb, Super Saver, Big Brown, Funny Cide, War Emblem, Fusaichi Pegasus, Charismatic, Silver Charm, and Lil E Tee.No Smarty Jones? I thought he qualified on this angle.

[SecretAgentMan]

No Smarty Jones? I thought he qualified on this angle.



Smarty had the 3rd highest beyer in his last prep before the derby.......The Cliffs Edge (111)......Lion Heart (110).......Smarty (109)

[rastajenk]

Forgot about Cliffy. I did remember that a 3-horse exacta box of high Beyers hit, but I thought Smarty was the leader of that pack.

[BlueChip@DRF]

Forgot about Cliffy. I did remember that a 3-horse exacta box of high Beyers hit, but I thought Smarty was the leader of that pack.

It was a superfecta that year when all last triple-digit beyers went 1-2-3-4.

I remember hearing the guy at the OTB saying he played an 4-horse exacta box using those four highest BSF.

[plainolebill]

If I remember correctly, Smarty's fig for the Arkansas Derby was adjusted upward - after the Kentucky Derby was run.

[dilanesp]

The sample size is statistically small meaning there is not gonna be a high degree of confidence that 40% figure is representative of the condition.

This. Remember just about any Derby handicapping angle has huge sample size problems. It's not like finding an angle in two turn mile allowance races or something. There is one Derby a year and racing luck looms huge.

[sbcaris]

Dilanesp: There is a test called the Fisher Test which can at least tell us whether or not the data is statistically significant and therefore not likely be due to a chance happening. I have used this test from time to time to see if there is a level of confidence that my stats are really are significant. Here is a test that can be performed on the data above.

If you type in Google --2 X 2 contingency table you will get a Fisher test with 4 rectangles as follows:

Observed Winners that Qualify-----10---------Expected Winners Qualified 3
Observed Winners Not Qualified----13--------Expected Winners Not Qual 20

Then all you need do is click on the calculate below and you will get the p value for the above which turns out to be .0472. A p value under .05 indicates that the data is statistically significant.

Although my data from 23 years may be considered a small sample by many statisticians, it is statistically significant.

One might ask how I got 3 for the expected winners. That number 3 comes from taking 14% of 23 years. Similarly, the 29 comes from taking 86% of 20 years.

[TexasDolly]

Dilanesp: There is a test called the Fisher Test which can at least tell us whether or not the data is statistically significant and therefore not likely be due to a chance happening. I have used this test from time to time to see if there is a level of confidence that my stats are really are significant. Here is a test that can be performed on the data above.

If you type in Google --2 X 2 contingency table you will get a Fisher test with 4 rectangles as follows:

Observed Winners that Qualify-----10---------Expected Winners Qualified 3
Observed Winners Not Qualified----13--------Expected Winners Not Qual 20

Then all you need do is click on the calculate below and you will get the p value for the above which turns out to be .0472. A p value under .05 indicates that the data is statistically significant.

Although my data from 23 years may be considered a small sample by many statisticians, it is statistically significant.

One might ask how I got 3 for the expected winners. That number 3 comes from taking 14% of 23 years. Similarly, the 29 comes from taking 86% of 20 years.

Hi SB, I wondered if you had mistyped that last sentence ?
I thought you might have meant 20 comes from taking 86% of 23.

One other question, if the chance of one of those horses winning the race is calculated using TM Bayes reasoning ,what would that chance be?

Thank you,
TD

[sbcaris]

Texas Dolly: Yes that was my error in typing and should be 20 instead of 29.

I do not know what ™ Bayes reasoning involves but based on the statistic that 10 of the last 23 fit this angle there should be a 43.5% chance that this years Derby winner would fit the angle.

[dilanesp]

Dilanesp: There is a test called the Fisher Test which can at least tell us whether or not the data is statistically significant and therefore not likely be due to a chance happening. I have used this test from time to time to see if there is a level of confidence that my stats are really are significant. Here is a test that can be performed on the data above.

If you type in Google --2 X 2 contingency table you will get a Fisher test with 4 rectangles as follows:

Observed Winners that Qualify-----10---------Expected Winners Qualified 3
Observed Winners Not Qualified----13--------Expected Winners Not Qual 20

Then all you need do is click on the calculate below and you will get the p value for the above which turns out to be .0472. A p value under .05 indicates that the data is statistically significant.

Although my data from 23 years may be considered a small sample by many statisticians, it is statistically significant.

One might ask how I got 3 for the expected winners. That number 3 comes from taking 14% of 23 years. Similarly, the 29 comes from taking 86% of 20 years.

There's an old saying in economics that people find it convenient to dismiss as unimportant anything that is difficult to measure....

You haven't given any reason to dismiss the need for a large sample and small confidence interval, other than you don't like the fact that it is not available and you would like to bet the Kentucky Derby.

Would a peer reviewed academic journal publish your statistical analysis of the Derby? Or would they say "too small a sample"?

EDIT: the Fisher test requires NOMINAL variables- on or off conditions such as sex (male or female).

The results of a horse race are an ordinal variable if you are studying only finish position (1st, 2nd, 3rd), or a measurement variable if you are studying performance (such as speed figures). The Fisher test does not measure statistical correlation in these circumstances.

[sbcaris]

Dilanesp: Maybe you can find a better angle using Beyer figs over the last 23 years (1992-2014) for use in the Kentucky Derby. When you do please post it here for all our forum members to see.

All I am interested in is getting winners of the Triple Crown races and the stats I have researched helped me do that on many occasions. For example---

Last year the stats I came up with on the Belmont Stakes led me to recommend using Tonalist as one of my keys in the exacta, trifecta and superfecta pools and that produced a huge cash return of over $4,500. Just a simple stat like Buck in the X plus final fraction time was a big part of that major hit. The winner, Tonalist, the place horse Commissioner, one fourth place finisher California Chrome and the other who dead heated with him, Wicked Strong all carried Buckpasser in the X and qualified on my final fraction indicator.

Furthermore, I recommended the winning exacta, the winning trifecta key, and the winning superfecta keys to all my customers at American Turf Monthly. If they followed my recommendations on the above exotic wagers they too would have hit the Belmont for over $4,500.

You might not be convinced that having Buckpasser in the X plus fast final fractions is a solid method for handicapping the Derby, Preakness and Belmont stakes because of the small sample size. However, that angle works like dynamite in the Derby, Preakness and Belmont stakes and I have illustrated the value of it in my book-Analyzing The Triple Crown.

Although the sample size is small, the method works like dynamite in all three Triple Crown races.

[Cratos]

Dilanesp: There is a test called the Fisher Test which can at least tell us whether or not the data is statistically significant and therefore not likely be due to a chance happening. I have used this test from time to time to see if there is a level of confidence that my stats are really are significant. Here is a test that can be performed on the data above.

If you type in Google --2 X 2 contingency table you will get a Fisher test with 4 rectangles as follows:

Observed Winners that Qualify-----10---------Expected Winners Qualified 3
Observed Winners Not Qualified----13--------Expected Winners Not Qual 20

Then all you need do is click on the calculate below and you will get the p value for the above which turns out to be .0472. A p value under .05 indicates that the data is statistically significant.

Although my data from 23 years may be considered a small sample by many statisticians, it is statistically significant.

One might ask how I got 3 for the expected winners. That number 3 comes from taking 14% of 23 years. Similarly, the 29 comes from taking 86% of 20 years.
Statistically significance from the Fisher's exact test of independence means that your number 3 is reliable; it doesn’t mean that it is important. To go into the importance of significance testing you might want to do what is called the t-test significance.

[dilanesp]

Dilanesp: Maybe you can find a better angle using Beyer figs over the last 23 years (1992-2014) for use in the Kentucky Derby. When you do please post it here for all our forum members to see.

All I am interested in is getting winners of the Triple Crown races and the stats I have researched helped me do that on many occasions. For example---

Last year the stats I came up with on the Belmont Stakes led me to recommend using Tonalist as one of my keys in the exacta, trifecta and superfecta pools and that produced a huge cash return of over $4,500. Just a simple stat like Buck in the X plus final fraction time was a big part of that major hit. The winner, Tonalist, the place horse Commissioner, one fourth place finisher California Chrome and the other who dead heated with him, Wicked Strong all carried Buckpasser in the X and qualified on my final fraction indicator.

Furthermore, I recommended the winning exacta, the winning trifecta key, and the winning superfecta keys to all my customers at American Turf Monthly. If they followed my recommendations on the above exotic wagers they too would have hit the Belmont for over $4,500.

You might not be convinced that having Buckpasser in the X plus fast final fractions is a solid method for handicapping the Derby, Preakness and Belmont stakes because of the small sample size. However, that angle works like dynamite in the Derby, Preakness and Belmont stakes and I have illustrated the value of it in my book-Analyzing The Triple Crown.

Although the sample size is small, the method works like dynamite in all three Triple Crown races.

Congratulations on your winning plays. I played poker last night, had ace ten on an A733 board, and was called by two players holding 97 and 76. They split the pot when the last 7 in the deck hit the river, and I lost. That does not, of course, mean that their plays were correct. Random variance is far more responsible for the success of one play than anything else, in both poker and horse racing.

In my lifetime, I've seen a lot of angles on the Kentucky Derby, many with a lot more than 23 years of data supporting them, blow up. Geldings, horses with less than 6 starts, dosage and the experimental free handicap, etc. The reason was sample size.

I'm going to give you some actual statistical guidance here. The problem with your model is that it treats winning as an on-off switch-- you either win or you don't. But 2nd place counts as a non-winner (so Cavonnier was a non-winner when he lost by a nose to Grindstone) and counts exactly the same as 20th place. But that's not right-- obviously, a horse that loses the Derby by a nose has delivered a far different performance than a horse who finished last. You are treating these as equivalent only so you can justify using the Fisher test.

If you want to use nominal variables, you need to treat each Derby as a set of match races. So the first horse in a 20 horse field has won 19 match races, the second horse has won 18 and lost 1, etc. So you then want to correlate any handicapping angle with the number of match races that horses win, and see if you have a statistically significant correlation.

The other way to do it is to use ORDINAL variables, but not use the Fisher test. You can treat finish position in the Derby as a ranking, create a data set, and then use a different test for significance. Some googling will point you in the right direction here.

But your current methodology doesn't prove anything, and saying "I won money with this" doesn't prove it either, because you haven't proven you can systematically outrun variance with it.

[rastajenk]

Wow...tough crowd. Second place may count as a non-winner, but it completes an exacta. :p

[Cratos]

Congratulations on your winning plays. I played poker last night, had ace ten on an A733 board, and was called by two players holding 97 and 76. They split the pot when the last 7 in the deck hit the river, and I lost. That does not, of course, mean that their plays were correct. Random variance is far more responsible for the success of one play than anything else, in both poker and horse racing.

In my lifetime, I've seen a lot of angles on the Kentucky Derby, many with a lot more than 23 years of data supporting them, blow up. Geldings, horses with less than 6 starts, dosage and the experimental free handicap, etc. The reason was sample size.

I'm going to give you some actual statistical guidance here. The problem with your model is that it treats winning as an on-off switch-- you either win or you don't. But 2nd place counts as a non-winner (so Cavonnier was a non-winner when he lost by a nose to Grindstone) and counts exactly the same as 20th place. But that's not right-- obviously, a horse that loses the Derby by a nose has delivered a far different performance than a horse who finished last. You are treating these as equivalent only so you can justify using the Fisher test.

If you want to use nominal variables, you need to treat each Derby as a set of match races. So the first horse in a 20 horse field has won 19 match races, the second horse has won 18 and lost 1, etc. So you then want to correlate any handicapping angle with the number of match races that horses win, and see if you have a statistically significant correlation.

The other way to do it is to use ORDINAL variables, but not use the Fisher test. You can treat finish position in the Derby as a ranking, create a data set, and then use a different test for significance. Some googling will point you in the right direction here.

But your current methodology doesn't prove anything, and saying "I won money with this" doesn't prove it either, because you haven't proven you can systematically outrun variance with it.
A very good response

[sbcaris]

Dilanesp: You said, " my current method does not prove anything."

Well if that is the case how do you explain why the method works so well over the history of the Derby, Preakness and Belmont Stakes? Is it just a chance occurrence that horses with Buckpasser in the X passing position plus a fast final fraction have won the Belmont Stakes 9 times in the last 20 years when at least one qualifier started in the race, have run second 7 times, have run third 3 times and have run fourth four times? Is it just chance that caused these 23 of 35 qualifiers to finish in the superfecta of the Belmont. Thats 65% of my qualifiers in the superfecta. Is it just due to chance that there were 45% of the Belmont winners in the last 20 years when at least one qualifier started from only 17% of the starters? Is it just due to chance that the impact value of this method is a strong 2.66? Is it just chance that caused 35% of these qualifiers to run second in the Belmont Stakes with a place impact value over 2.00.

[dilanesp]

Dilanesp: You said, " my current method does not prove anything."

Well if that is the case how do you explain why the method works so well over the history of the Derby, Preakness and Belmont Stakes? Is it just a chance occurrence that horses with Buckpasser in the X passing position plus a fast final fraction have won the Belmont Stakes 9 times in the last 20 years when at least one qualifier started in the race, have run second 7 times, have run third 3 times and have run fourth four times? Is it just chance that caused these 23 of 35 qualifiers to finish in the superfecta of the Belmont. Thats 65% of my qualifiers in the superfecta. Is it just due to chance that there were 45% of the Belmont winners in the last 20 years when at least one qualifier started from only 17% of the starters? Is it just due to chance that the impact value of this method is a strong 2.66? Is it just chance that caused 35% of these qualifiers to run second in the Belmont Stakes with a place impact value over 2.00.

Was it a chance occurrence that no gelding won the Derby since Clyde Van Deusen, until Funny Cide did it? (BTW, I remember debating this online with a pretty decent handicapper back in 2001 or 2002 who was convinced the gelding thing was real and had all sorts of theories as to why a gelding couldn't stand the stress of the big field on Derby day or whatever.) Was it a chance occurrence that no Derby winner for 40 years had a Dosage over 4.0, until Strike the Gold won?

I mean, yeah, you are dealing with a freaky race, there's a ton of statistical noise and variance, and it's possible to cash lots of tickets without actually having identified a real correlation.

I'm happy you are winning money. I hope it continues for you. But that doesn't mean that you've proven the relationship-- the next 20 Derbies may play out completely differently, or maybe they won't.

All I can do is point you in the right direction as to the type of statistical model you really need to construct to try and prove this, and I have done that. The rest is a matter of hard work, careful modeling, and time necessary to outrun the variance....

[Cratos]

Was it a chance occurrence that no gelding won the Derby since Clyde Van Deusen, until Funny Cide did it? (BTW, I remember debating this online with a pretty decent handicapper back in 2001 or 2002 who was convinced the gelding thing was real and had all sorts of theories as to why a gelding couldn't stand the stress of the big field on Derby day or whatever.) Was it a chance occurrence that no Derby winner for 40 years had a Dosage over 4.0, until Strike the Gold won?

I mean, yeah, you are dealing with a freaky race, there's a ton of statistical noise and variance, and it's possible to cash lots of tickets without actually having identified a real correlation.

I'm happy you are winning money. I hope it continues for you. But that doesn't mean that you've proven the relationship-- the next 20 Derbies may play out completely differently, or maybe they won't.

All I can do is point you in the right direction as to the type of statistical model you really need to construct to try and prove this, and I have done that. The rest is a matter of hard work, careful modeling, and time necessary to outrun the variance....
How right you are

[sbcaris]

dilanesp: I gave some strong stats in my last post that point to the effectiveness of my Buckpasser in the X position angle on the Belmont stakes. It should be noted that a similar angle using Buckpasser in the X with fast final fractions has strong impact values and ROIs in the Derby and Preakness.

If it were just one race of the three where it worked to get winners and in the money finishers, I would agree with you that it might just be due to chance but in all three Triple Crown events my Buck in the X angle comes up strong with regard to impact values and ROIs.

One more important fact that supports my research. My book -Analyzing The Triple Crown was published by American Turf Monthly in March of 2013. In that book I illustrated by means of statistics the importance of Buckpasser in the X passing position in all three legs of the Triple Crown.

A couple of months after its publication Orb, the only horse in the Derby with Buckpasser in the X passing position who qualified on my final fraction indicator won the roses. Then in 2014 California Chrome, another qualifier on my Buck in the X angle wins the roses. Then in the Preakness the exacta is comprised of two horses that qualified on this angle: California Chrome and Ride On Curlin. Then in the Belmont 4 of the 5 horses that comprised the superfecta were filled by horses that fit my Buck in the X angle: Tonalist, Commissioner, California Chrome and Wicked Strong.

Let me point you in the right direction, my Buckpasser in the X position plus fast final fractions works like a charm.

[SecretAgentMan]

Hey sbcaris, do we have any Buckpass in the X plus the final fraction horses this year?

I thought I saw there isn't any qualifiers so far in this KD.......

[MJC922]

The problem with the derby and angle 'fitting' is always sample size. no disrespect to the OP, I still love to read the interesting posts even if I know with regard to the Beyer fig it's likely to all equal out eventually, maybe not until we're all long gone but it will equal out.

What you're looking at is selective application of the Beyer figure, which someone correct me if I'm wrong win% is in the upper 20's. Non-selective application, top fig without regard to distance and surface is 25% from sport stat. If it's hitting 40% that's great but it's just on a good run for now and will come back down someday.

It's a great race which generates massive interest but still, let's not forget the law of large numbers. It's only one race per year. All kinds of interesting back testing that can be done to 'fit' and create cutoff values etc, but no way to forward test any of it. I can create dozens of black boxes that bet every race and get positive ROIs, none of it holds up in forward testing though unless I start to get ultra selective and even then we're talking very low +ROI.

[sbcaris]

SAM: Not yet. This saturday three may go in the Ark Derby: The Truth or Else, Bold Conquest, and Madefromlucky. If none of these run a fast final fraction it will be one of those years with no qualifiers. I will use other angles before making my final decisions.

Its a powerful angle but qualifiers only occur in around 50% of the Derbies all the way back to 1978 when the very first runner qualified- Believe It who ran third that year to Affirmed and Alydar.

[BlueChip@DRF]

Does one more generation back make that much of a huge difference?

[sbcaris]

Bluechip@DRF: Did you ask me that question? If so what exactly do you mean by one more generation back?

[BlueChip@DRF]

Bluechip@DRF: Did you ask me that question? If so what exactly do you mean by one more generation back?


You stated the Buckpasser X had to be within 5 generations. What if he was in a primary passing position in the 6th generation?

[sbcaris]

Bluechip@drf: My angle requires that Buckpasser has to be in the X passing position. I never said he has to be within 5 generations.

[OCF]

Was it a chance occurrence that no gelding won the Derby since Clyde Van Deusen, until Funny Cide did it? (BTW, I remember debating this online with a pretty decent handicapper back in 2001 or 2002 who was convinced the gelding thing was real and had all sorts of theories as to why a gelding couldn't stand the stress of the big field on Derby day or whatever.) Was it a chance occurrence that no Derby winner for 40 years had a Dosage over 4.0, until Strike the Gold won?

I mean, yeah, you are dealing with a freaky race, there's a ton of statistical noise and variance, and it's possible to cash lots of tickets without actually having identified a real correlation.

I'm happy you are winning money. I hope it continues for you. But that doesn't mean that you've proven the relationship-- the next 20 Derbies may play out completely differently, or maybe they won't.

All I can do is point you in the right direction as to the type of statistical model you really need to construct to try and prove this, and I have done that. The rest is a matter of hard work, careful modeling, and time necessary to outrun the variance....

Yes, sbcaris's sample size is relatively small. But since we'd probably all agree that that is an inescapable limitation to trying to model the KD, IMHO it seems that we either live with that or go with what amounts to applying the "eye test" anew to every horse's race replays, every horse's PP's, every year.

Given sbcaris's results I'm leaning heavily towards his model, even with the limitations.

I guess what I'm saying is his results pass my eye test. ;)