Reverse Impact Values?

[podonne]

In my continued quest to pick the loser, I wonder if an Impact Value might be equally as useful to indicate which horse won't win a race. For instance, an impact value of 0.10 might indicate that a certain group is only 10% as likely to win a race as random chance.

Emperically it seems to make sense, but I am don't have a mastery of impact value thoery, so I thought I would ask those who are.

Philip S. O'Donnell

[Tom]

I use IV's more as an indicator of unlikely success than I do as a postive sign,
like you propose here.

[twindouble]

I use IV's more as an indicator of unlikely success than I do as a postive sign,
like you propose here.

I don't know what you two are getting at but handicappers are supposed to look at a horse's chance of winning under the conditions he's entered. In a given race his chances could be very high even if he's 30-1 or more. Why give a horse a number based on stats that has nothing to do with the race he's in now? Please explain.


T.D.

[Overlay]

I understand what you're saying about using impact values as a criterion for eliminating horses, and (to me) low impact values would be more reliable than non-quantitative standards. I've tended to shy away from that, however, based on the belief that the horses you toss will include the ones that will jump up and surprise you at high mutuels. I realize that any sound, long-term approach has to focus on horses at the lower ends of the odds spectrum, and that there's no point or profit in betting high-odds horses indiscriminately trying to get lucky. What's worked for me is viewing tools like impact values not as a means of eliminating horses, but as a standard for ranking and assigning a meaningful winning probability to each of them (including both "contenders" and "non-contenders"), based not just on one factor or angle, but on a combination of fundamental elements. That way, I can spot and quantify betting value wherever it might fall within the odds spectrum, and size my wager according to the degree of risk involved.

[Tom]

I don't know what you two are getting at but handicappers are supposed to look at a horse's chance of winning under the conditions he's entered. In a given race his chances could be very high even if he's 30-1 or more. Why give a horse a number based on stats that has nothing to do with the race he's in now? Please explain.


T.D.

The IV is based on the situation today.

[twindouble]

The IV is based on the situation today.

Thanks for not shooting me.:D

That makes sense but what is "IV". Don't keep it a secret.


T.D.

[Overlay]

Impact Value (the percentage of race winners possessing a given handicapping attribute divided by the percentage of all horses with that same attribute). The higher the IV for a factor, the greater its power in predicting or influencing a horse's chance of winning. (Popularized by William Quirin in Winning at the Races, but in use before that by people such as Frederick Davis.)

[CBedo]

I often use low impact values to help make logical eliminations. Using a spreadsheet that uses impact values or a program like All-Ways, If I have a contender selection that uses impact values to select the "best" horses, I will often add one more selection criteria where any horse that falls in the bottom half (or sometimes as tight as not in the top two or three) shows a terrible impact value. Instead of using the calculated impact values, I will arbitrarily assign a value of 1.0 to the top half (or top two or three) and usually around 0.5-0.7 to the rest. This doesn't automatically eliminate them, but it means they have to have spectacular values in any of the other criteria to keep from being eliminated.

[podonne]

Thanks for the comments, from I gather it is mathmatically sound. I'm going to post a new thread about the math of combining IVs.

[betovernetcapper]

Quirin looked at the number of failures in a horses record to do what he's trying to do today (example sprint nw-2 claiming) and found

0 failures IV 1.23
1 failure IV .79
2+ failures IV .49

Horses with repeated failures really seem to underpreform.

Quirin also found that horses with no speed points had an IV of .65. Depending on the track distance that IV is can be a lot lower. On the same lines, horses that haven't won or had the lead at any point of call in their last 10 races don't win very often, duh, but sometimes are still bet.

[classhandicapper]

On the same lines, horses that haven't won or had the lead at any point of call in their last 10 races don't win very often, duh, but sometimes are still bet.

Things like that can sometimes get very tricky.

Suppose the horse is dropping in class from a MSW to a Statebred Maiden Claimer.

A horse's poor record could reflect the level of competition and not it's own lack of competitive spirit and ability. It's tough to automate the analysis...at least for me.

[betovernetcapper]

The non-winner last 10 combined with no leads at ANY point of call last 10 is an unlikely winner no matter what the class. If it's been unable to win or get the lead at any point in it's last 10, I wouldn't credit it as having the class of the horses it's been running with. :)

[podonne]

Ah, but what if the last 10 races were G1 stakes where he finished 2nd and the horse is entered in a Clm5000? True, probably won't happen often, but illustrates the danger with making generalized statements. I ran into this problem alot when trying to come up with good non-contender evaluations. Almost impossible. :)

[betovernetcapper]

I'll be happy to accept all bets on on horses droping from a placing in a graded race to $5000 claiming. :)