STATISTICS: When is a 3rd Finish like a 1st?

[Triopstor]

Hi.

I hope someone with a database might be able to answer this question.


BACKGROUND
-------------
William L. Quirin mentions that a Finish of 1st in the last race = 17.0% win percentage, impact value of 1.52. While a Finish of 2nd in the last race = 17.6% win percentage, impact value 1.58. And a Finish of 3rd in the last race = 13.0% win percentage, impact value 1.17.

Keep in mind the horse could have last ran 1st/2nd/3rd several months ago. Date is not a consideration.


THE QUESTION
---------------

So how can the handicapper raise the lowly 3rd Finish horse to a level of a 1st Finish horse of winning 17.0%?

What if we could use a database and find that horses that Finished 3rd but within 3/4 of a length wins 17.0% of the next race?

What if a Finish of 4th but within 1/2 a length wins next time 17.0% of the time?

What if a Finish of 5th but within a nose(ns) wins next time 17.0% of the time?

And so on.

This way if a selection system has a rule of "1. Horse must have finished 1st or 2nd last race" that we can expand this to include other types of horses that didnt' Finish 1st or 2nd last race? Should work but maybe not.

[ranchwest]

I can understand that someone with a database might answer a question for you, but you are asking for the question. I doubt if anyone is going to tell you the question.

[Overlay]

Quirin based many of his findings/values on when a given horse last ran what he called a "good race", which he defined as a finish of first, second, or third (regardless of how many lengths by which the horse finished ahead or behind), or (even if the horse did not finish in the top three positions), a finish within two lengths of the winner in a sprint (a race of less than one mile), or within three lengths of the winner in a route (a race of one mile or more). (I'm not sure what his statistical basis was for that definition, but that is what he used.) So, for purposes of working with his statistics related to the horse's last "good race" (which were fairly extensive, including as part of his regression equation for sprints), a third-place finish was/would be statistically indistinguishable from a first-place finish.

Thus, from that standpoint, if you're using Quirin as your frame of reference, I'm not sure that it pays to do too much hair-splitting about the comparative value of a third-place finish versus a first-place finish, or to get down into the weeds of distinguishing between finishes on the basis of lengths behind the winner.

[formula_2002]

What if we could use a database and find that horses that Finished 3rd but within 3/4 of a length wins 17.0% of the next race?.
It would be even more usefull to know the dollar odds of the 3rd place finisher.
IMO, when doing these "What Ifs", you want to know the dollar odds of the horses. In the long run, a horse's actual win pool wins/(1/odds+1) adjusted for track takeout ratio increases with the shorter odds.
To find a horses probability of finishing 3rd, I would use a data base of a few thound races to find win odds, win pool a/e, 3rd finishes.
In the long run, the win pool odds should adjust to all racing conditions, i.e. track, distance,pace,etc.. of course, you would want to verify even that with a very large data base. :)

[46zilzal]

When one overlays the yardstick of PACE OF RACE a horse battled against, POSTION is irrelevant.

How the horse did against the PACE not the position is the important thing. I see horses win all the time from pace lines where they came in 7th or 8th repeatedly.

PACE of RACE, not position which is only RELATIVE tho how the OTHERS did that day, not the runner itself.

[Triopstor]

I can understand that someone with a database might answer a question for you, but you are asking for the question. I doubt if anyone is going to tell you the question.

I'm asking for an answer yes. Looks like a job for JCapper! You doubt? Why should you doubt among friends?

Sincerely Triopstor.

[Triopstor]

Quirin based many of his findings/values on when a given horse last ran what he called a "good race", which he defined as a finish of first, second, or third (regardless of how many lengths by which the horse finished ahead or behind), or (even if the horse did not finish in the top three positions), a finish within two lengths of the winner in a sprint (a race of less than one mile), or within three lengths of the winner in a route (a race of one mile or more). (I'm not sure what his statistical basis was for that definition, but that is what he used.) So, for purposes of working with his statistics related to the horse's last "good race" (which were fairly extensive, including as part of his regression equation for sprints), a third-place finish was/would be statistically indistinguishable from a first-place finish.

Thus, from that standpoint, if you're using Quirin as your frame of reference, I'm not sure that it pays to do too much hair-splitting about the comparative value of a third-place finish versus a first-place finish, or to get down into the weeds of distinguishing between finishes on the basis of lengths behind the winner.

Yes I familiar with a Quirin's good race as my source was his book "Winning At The Races: Computer Discoveries In Handicapping 1979." Which was based upon previous research articles he did about 8-5 years prior I believe. Thus the book is a redaction.

What is the meaning behind "third-place finish was/would be statistically indistinguishable from a first-place finish"? I do not understand as I quoted some of Quirins statistics above in the main thread post.

I believe this detail is of importance such as man's invention of the microscope. New discoveries can be found looking into deeper matters. Don't recall any exploration of such kind ever being fruitless. Maybe looking deeper always has a value of love.

Sincerely Triopstor.

[Overlay]

What is the meaning behind "third-place finish was/would be statistically indistinguishable from a first-place finish"? I do not understand as I quoted some of Quirins statistics above in the main thread post.

I realize that Quirin included probabilities/impact values associated with a horse's specific finishing position in its latest race. However, the point that I was making was that he also based many of his findings on the concept of when the horse last ran a "good race" (as Quirin defined that term), without making distinctions between specific finishing positions or exact number of lengths ahead or behind, as long as the horse's finish met either the general position or lengths-behind criteria that Quirin stipulated.

Since a race would count as a "good race" for a horse whether it finished first, second, or third, the effect would be to treat first-place finishes and third-place finishes the same with respect to a particular angle or system where the concept of a "good race" is being used as a qualifying factor.

Is that any clearer?

[Overlay]

Let me just add that I recognize that not all horses that run a "good race" would be equal from the standpoint of the effect of that race upon their chance of winning a future race. However, since Quirin treated them as equal by using the "good race" concept, and did not provide a specific breakout of how many of the "good race" horses finished first, second, or third, or finished worse than third but qualified for a "good race" under the lengths-behind criterion, I don't see how you could "reverse-engineer" his findings related to "good races" to determine probabilities associated with the various individual finishing positions, or degrees of lengths ahead or behind, if you have only the aggregate "good race" statistics to work from.

[Triopstor]

It would be even more usefull to know the dollar odds of the 3rd place finisher.

IMO, when doing these "What Ifs", you want to know the dollar odds of the horses. In the long run, a horse's actual win pool wins/(1/odds+1) adjusted for track takeout ratio increases with the shorter odds.
To find a horses probability of finishing 3rd, I would use a data base of a few thound races to find win odds, win pool a/e, 3rd finishes.
In the long run, the win pool odds should adjust to all racing conditions, i.e. track, distance,pace,etc.. of course, you would want to verify even that with a very large data base. :)

Thanks for your constructive reply. What you propose is indeed very useful. Yes I remember there was a Jcapper posting about this but I can't find the thead and I accidently may have deleted this from the computer.

Sincerely Triopstor

[Triopstor]

When one overlays the yardstick of PACE OF RACE a horse battled against, POSTION is irrelevant.

How the horse did against the PACE not the position is the important thing. I see horses win all the time from pace lines where they came in 7th or 8th repeatedly.

PACE of RACE, not position which is only RELATIVE tho how the OTHERS did that day, not the runner itself.

Why not combine the two?


Sincerely Triopstor.

[LottaKash]

I would ask; was this 3rd place horse, Improving, Regressing, or Staying the Same in his current "Form-Cylcle", before that race was run ?...That would explain much...to me...

And of course, the "pace of the race" (especially a swift one) and his involvement vs. that pace.....Was he just a "suckalong" or a "true-contender" making some move or showing effort vs. that pace ?...

I believe that these questions must be addressed, otherwise finishing 3d is plain meaningless, at least to me...

best,

[Triopstor]

I would ask; was this 3rd place horse, Improving, Regressing, or Staying the Same in his current "Form-Cylcle", before that race was run ?

Maybe very powerful what you propose. Those questions can be asked of any horse including those that finished 1st or 2nd last race outing.

Knowing how close the horse finished in lengths may decrease the handicapping process of those questions you asked. The data may overlap therefore making the task extreneous to look at other angles of perception.

But we won't know till someone provides the keys to test all the locked doors.

Thank you very much for your reply.