Race condition that is least chalky

[coachgonzo]

I am appealing to those with the large data bases for a more accurate picture.

It is my understanding that other than graded races (I & II, especially), maiden races are the chalkiest.


Is there a condition that is least likely to produce a chalky victor? MSW; MDCL; nw2,3; nw1x,2x; cheap open claimers; higher end open claimers; nw of races in certain dated periods; etc.

Any guidance would be greatly appreciated.

Thanks

[Overlay]

My data shows maiden turf routes for three-year-olds and up as a specific subset in terms of both low percentage of winning favorites and high average mutuel, and turf routes in general as a broader category.

[RXB]

Field size probably has a lot to do with that.

[46zilzal]

Each race is distinct unto itself and to compare a RACE with the crowd's wagering is akin to comparing ice on your winter driveway as a function of yogurt prices at the market.

[Greyfox]

Each race is distinct unto itself and to compare a RACE with the crowd's wagering is akin to comparing ice on your winter driveway as a function of yogurt prices at the market.

Gosh darn another one who can't read or comprehend.
I think most of us find each race distinct unto itself. That is not a breakthrough observation.
The thread starter is requesting what condition, in general,offers the best betting value?

[46zilzal]

correlation does not always have anything whatsoever to do with cause and effect....NEVER HAS

This constant idea of breaking things down into components and then somehow finding these correlations is, at times, laughable.

[Greyfox]

My data shows maiden turf routes for three-year-olds and up as a specific subset in terms of both low percentage of winning favorites and high average mutuel, and turf routes in general as a broader category.

Overlay,

I don't do data base work but intuitively I'd support your observations here.
My suspicion is that NW 2 for cheap claimers also have low percentage winning favorites and high average mutual.
Does your data base crystal ball see any support for that view??

[Bettowin]

Not really a condition but I see lots of long odds early in new meets. The first 1-3 weeks especially at tracks that aren't in close proximity to others so you see horse coming from many different tracks and off layoffs.

[Overlay]

I don't do data base work but intuitively I'd support your observations here.
My suspicion is that NW 2 for cheap claimers also have low percentage winning favorites and high average mutual.
Does your data base crystal ball see any support for that view??

You may very well be correct, since it stands to reason that the lower the class of a race is, and the less impressive the credentials of the entrants are in terms of wins or earnings, the more wide open/unpredictable the race would be, but my data is not broken down that fine (nor do I have the means to do so). Sorry.

[sjk]

type win pct field size
MSW and Lifetime NW Allow 37.68% 8.20
Open Claiming 34.74% 7.96
Maiden Claiming 36.68% 8.86
Open Allowance and Stk 40.16% 7.55
Lifetime NW Claiming 34.81% 8.47


Sorry about formatting but hope the content is clear.

[Grits]

Not really a condition but I see lots of long odds early in new meets. The first 1-3 weeks especially at tracks that aren't in close proximity to others so you see horse coming from many different tracks and off layoffs.

I agree. This is certainly true of Saratoga with so many horses coming in from tracks all over, not just downstate.

[ranchwest]

Gosh darn another one who can't read or comprehend.
I think most of us find each race distinct unto itself. That is not a breakthrough observation.
The thread starter is requesting what condition, in general,offers the best betting value?

Actually, the question was about price, not value.

[Zaf]

I don't have a database, but I would guess those cheap 14 horse fields of Maiden Claimers or NW1Y at EVD should fit the bill. :)

Z

[Greyfox]

Actually, the question was about price, not value.

How silly of me. Unfortunately I used a synonym dictionary that says that
Price = monetary value.

http://www.synonyms.net/synonym/price




:cool:

[ranchwest]

How silly of me. Unfortunately I used a synonym dictionary that says that
Price = monetary value.

http://www.synonyms.net/synonym/price




:cool:

I'm glad you think that. Puts more money in the pools for the rest of us. Thanks.

[Greyfox]

I'm glad you think that. Puts more money in the pools for the rest of us. Thanks.

Why not? It wouldn't be the first time I subsidized farmers. :lol:

[gm10]

You want to be looking at Impact Values. They take field size into account. I've posted a superficial IV analysis wrt the Pro-Ride @ SA last year.

[skate]

if i had to pic a condition for "least chalk (REAL CHALK)id call for a turf race that had NO previous turf horses.


which might have a 4/5 horse, but it would be a very LOW (expected) win rate for 4/5.

[Sham]

From my HSH database...

Year 2009
Dirt and Poly, no Turf
All ages, both genders, all distances, all tracks (except obscure fair meets)

Public Choice 1 Impact Value

All Races IV=2.79
Claiming -all IV=2.70
Claiming -nw2L IV=2.86
Alw -NW1x IV=2.67
Alw -NW2x IV=2.57
Classified Alw IV=2.26
Starter Alw IV=2.54
Hdcp+Stakes IV=2.79
Graded IV=2.66
MSW IV=3.15
Mdn Clm IV=3.08

[gm10]

From my HSH database...

Year 2009
Dirt and Poly, no Turf
All ages, both genders, all distances, all tracks (except obscure fair meets)

Public Choice 1 Impact Value

All Races IV=2.79
Claiming -all IV=2.70
Claiming -nw2L IV=2.86
Alw -NW1x IV=2.67
Alw -NW2x IV=2.57
Classified Alw IV=2.26
Starter Alw IV=2.54
Hdcp+Stakes IV=2.79
Graded IV=2.66
MSW IV=3.15
Mdn Clm IV=3.08

Are these the IV's for the public favourites???

[Sham]

Are these the IV's for the public favourites???

Correct

Public Choice 1 Impact Value

[gm10]

Correct

I am very surprised. How did you define IV?

[Overlay]

I am very surprised. How did you define IV?

I assume that it would be the percentage of all winners of the races in the sample that were favorites, divided by the percentage of all horses in the sample that were favorites.

[Bill Cullen]

correlation does not always have anything whatsoever to do with cause and effect....NEVER HAS

This constant idea of breaking things down into components and then somehow finding these correlations is, at times, laughable.

Maybe a better way to say it is that correlation is a necessary condition but not sufficient enough by itself to guarantee causality.

Bill C

[46zilzal]

Maybe a better way to say it is that correlation is a necessary condition but not sufficient enough by itself to guarantee causality.

Bill C
I can buy that.

[gm10]

I assume that it would be the percentage of all winners of the races in the sample that were favorites, divided by the percentage of all horses in the sample that were favorites.

how could that number ever be bigger than 1?

[gm10]

IV is (by my definition anyway) the number of actual winners divided by the number of expected winners

the IV of favourites would be around 1 for most categories
to see an IV of 2 and even 3 makes me think we are not talking about the same thing here

[Overlay]

how could that number ever be bigger than 1?

The impact value for any particular factor will be greater than 1.00 whenever the percentage of race winners exhibiting that factor is greater than the percentage of all horses exhibiting it.

For example (to cite a case that Bill Quirin used to explain the impact value concept), in a 4800-race sample, there were a total of 4,800 favorites. Those favorites represented 10.9% of all the horses that ran in those 4,800 races. If the favorites won at a rate corresponding to their "fair share" of races, they would win 10.9% of the races, and the impact value for favorites in the sample would be 10.9% divided by 10.9%, or 1.00.

However, in fact, the favorites won 1,567 of the 4800 races, or 32.6% of all the races in the sample. Thus, they won 32.6/10.9, or 2.99 times their "fair share". This 2.99 impact value indicates that being a favorite is a factor that correlates positively with winning probability, with the correlation being stronger as the impact value gets further above 1.00.

By contrast, factors with impact values below 1.00 are negative indicators, with horses possessing those factors winning less and less of their "fair share" of races as the impact values for the factors approach zero.

Of course, a high impact value by itself does not guarantee profitability, as the example with favorites illustrates. The $NET from betting each favorite in Quirin's sample resulted in an overall loss of 9.5%.

[gm10]

The impact value for any particular factor will be greater than 1.00 whenever the percentage of race winners exhibiting that factor is greater than the percentage of all horses exhibiting it.

For example (to cite a case that Bill Quirin used to explain the impact value concept), in a 4800-race sample, there were a total of 4,800 favorites. Those favorites represented 10.9% of all the horses that ran in those 4,800 races. If the favorites won at a rate corresponding to their "fair share" of races, they would win 10.9% of the races, and the impact value for favorites in the sample would be 10.9% divided by 10.9%, or 1.00.

However, in fact, the favorites won 1,567 of the 4800 races, or 32.6% of all the races in the sample. Thus, they won 32.6/10.9, or 2.99 times their "fair share". This 2.99 impact value indicates that being a favorite is a factor that correlates positively with winning probability, with the correlation being stronger as the impact value gets further above 1.00.

By contrast, factors with impact values below 1.00 are negative indicators, with horses possessing those factors winning less and less of their "fair share" of races as the impact values for the factors approach zero.

Of course, a high impact value by itself does not guarantee profitability, as the example with favorites illustrates. The $NET from betting each favorite in Quirin's sample resulted in an overall loss of 9.5%.

Ah OK. We were using the same term for different things. I use Impact Value to measure how well horses do relative to their odds. Hypothetical example: category turf/route/maidens has a sample size of 1000. The average odds of the 1000 favourites are 2/1. In that case you would expect 333 favourites to win. Suppose only 200 favorites win. Then the Impact Value is 200/333 = 0.60. IV's greater than 1 mean that the public is underestimating the winning chances (and vice versa).

[Overlay]

Ah OK. We were using the same term for different things. I use Impact Value to measure how well horses do relative to their odds. Hypothetical example: category turf/route/maidens has a sample size of 1000. The average odds of the 1000 favourites are 2/1. In that case you would expect 333 favourites to win. Suppose only 200 favorites win. Then the Impact Value is 200/333 = 0.60. IV's greater than 1 mean that the public is underestimating the winning chances (and vice versa).

What you describe sounds similar to the way that Mike Nunamaker calculated impact values in his studies. The only problem that I had with using that approach was that, since all the other impact value data that I had was done Quirin's way, and I did not have access to odds data that would have allowed me to reconstruct those values using Nunamaker's method so that they could be directly compared, I recalculated all of Nunamaker's values using Quirin's formula, so that they would all have the same basis. (I don't recall the overall differences being that significant, with which I believe Nunamaker agreed when I asked him about it several years ago.)(I'd have to find the reply that he sent me to quote him directly.)

[Overlay]

Maybe a better way to say it is that correlation is a necessary condition but not sufficient enough by itself to guarantee causality.

Bill C

Bill,

The envelope arrived today. (Your PM box is full.)

[Overlay]

What you describe sounds similar to the way that Mike Nunamaker calculated impact values in his studies. The only problem that I had with using that approach was that, since all the other impact value data that I had was done Quirin's way, and I did not have access to odds data that would have allowed me to reconstruct those values using Nunamaker's method so that they could be directly compared, I recalculated all of Nunamaker's values using Quirin's formula, so that they would all have the same basis. (I don't recall the overall differences being that significant, with which I believe Nunamaker agreed when I asked him about it several years ago.)(I'd have to find the reply that he sent me to quote him directly.)

I stand corrected. I know that there was an author that I've read who incorporated odds into the calculation of impact values, but I don't believe that it was Nunamaker. The calculation that Nunamaker used took the field size for each of the individual races in the sample into consideration, rather than using the aggregate number of horses in the sample as a whole (as Quirin did). The end result is the same, however. Without knowing those field sizes, you can't directly compare impact values arrived at in that manner with impact values calculated as Quirin figured them, unless you recalculate them. However, again, the differences between them should not be that significant.

[Sham]

PIV, or Pool Impact Value (Pool Expected Winners/Actual Winners) is calculated in HSH as well...what I posted was simply IV of post time favorites.

[gm10]

What you describe sounds similar to the way that Mike Nunamaker calculated impact values in his studies. The only problem that I had with using that approach was that, since all the other impact value data that I had was done Quirin's way, and I did not have access to odds data that would have allowed me to reconstruct those values using Nunamaker's method so that they could be directly compared, I recalculated all of Nunamaker's values using Quirin's formula, so that they would all have the same basis. (I don't recall the overall differences being that significant, with which I believe Nunamaker agreed when I asked him about it several years ago.)(I'd have to find the reply that he sent me to quote him directly.)

I've certainly never discovered an IV of 3. I would be a very rich man if I had!!

[Tom]

Trouble with IV is it doens't take into account prices, which is what the thread started about. I would think the original question would best be answered by pecentage of non-favorites winning, or average payoff.

[valueguy]

According to Robert Rowe(old time handicapper) and my own observations
Field size is the most important varience when it comes to determing favorite
win percentage.I have been making my own value line for some time now ,and
the number of horses in the field determine at what price you will make your
favorite .I use only 3 for 4 horses for 80% of the line.
If you want a higher favorite price suggest you play fields in which there are
10 or more horses.FWIW