Horse stats vs ROI Relationship

[raybo]

Is there a consistent relationship between some kind of integration of (starts#, win#, place#, show#, ITM#, Total earnings) versus ROI%, assuming you are using a very large database of horse stats inclusive of all races other than first time starters (and possibly requiring at least 3 or 4 lifetime starts? In other words is there a fairly direct relationship between horses' racing experience, wps performance, total earnings .... and the win ROIs they produce?

It seems that I at least started a study of this, way back in my very early pen and paper analysis of results charts, but I'll be danged if I can remember how, or what I did, or what the results, if I actually went far enough to determine results, were. I'm sure many of you have traveled this road before and have found either a negative, positive, or neutral/random relationship, depending on how you integrated those factors.

Just some recent observations that triggered my curiosity. if I'm stepping on someone's profitable proprietary methods, I apologize.

By the way, please don't just tell me that a regression analysis of some kind would tell me the relationships. I don't have that mathematical/technical ability.

Hints at how to proceed are welcomed.

[Robert Goren]

The number of starts and total earning was part of system I developed from a multi regression study I did in 1973. The system showed a small profit until the late 1980s when horses stopped running as often as did before. The regression started with comparing odds to finish. The numbers of starts and total earnings increase the accuracy of that regression by a small amount. It generally picked the favorite and favorites picked by it showed a small profit, but when it picked a non favorite, it was more profitable. I will say this. The study was based on a very small sample of races from the Nebraska circuit and few from Sportsman Park thrown in. I was dealing with punch cards and the SPSS program on UNL IBM 370. My access to the computer was only 2 months long and I did not a chance to do every thing I want to do. Days away and distance switches were the other important factors in the system. I am sorry I do not have weights anymore.

[classhandicapper]

If I understand the question, Bill Quirin did some work along those lines using Win% and various measurements of consistency and class. That work is probably dated in terms of ROI.

[flatstats]

I only have data for UK flat racing so this may not be relevant to you.

In 3yo+ Handicap races there is a distinct cut-off at around 50 career runs.

Lower than that the A/E and IV figures are consistent. But from 50+ a horses' chance of winning is much more diminished.

Looking at the Horse Age figure 8yo and older seems to follow suit in that it is a threshold where ability declines.

Interestingly the Age IV is different from the A/E at less than 8.

Ages in 3yo+ Handicaps, UK Flat Turf 2004-2014
3yo 1.22 IV, 0.88 A/E
4yo 1.05 IV, 0.87 A/E
5yo 0.99 IV, 0.85 A/E
6yo 0.86 IV, 0.81 A/E
7yo 0.82 IV, 0.83 A/E
8yo+ 0.65 IV, 0.71 A/E

The IV is highest at 3yo and slopes down. But the A/E (based on actual over expected winners) remains consistent with just a small decline at 6yo and a big drop at 8yo+. This implies that the market tends to compensate for younger horses, who would tend to win more races.

[raybo]

I only have data for UK flat racing so this may not be relevant to you.

In 3yo+ Handicap races there is a distinct cut-off at around 50 career runs.

Lower than that the A/E and IV figures are consistent. But from 50+ a horses' chance of winning is much more diminished.

Looking at the Horse Age figure 8yo and older seems to follow suit in that it is a threshold where ability declines.

Interestingly the Age IV is different from the A/E at less than 8.

Ages in 3yo+ Handicaps, UK Flat Turf 2004-2014
3yo 1.22 IV, 0.88 A/E
4yo 1.05 IV, 0.87 A/E
5yo 0.99 IV, 0.85 A/E
6yo 0.86 IV, 0.81 A/E
7yo 0.82 IV, 0.83 A/E
8yo+ 0.65 IV, 0.71 A/E

The IV is highest at 3yo and slopes down. But the A/E (based on actual over expected winners) remains consistent with just a small decline at 6yo and a big drop at 8yo+. This implies that the market tends to compensate for younger horses, who would tend to win more races.

Thanks Flatstats! I was thinking that age and sex probably should be included in this study. Form (current physical condition) should probably be included also, as that factor appears to improve the results of any study, regardless of the factors involved.

[Dave Schwartz]

Ray,

I think you have my Percentages & Probabilities book. If so, look up the Earnings Box Class chapter.

If not, for now I will simply say that "consistency" produces highly disappointing results.

Absent some kind of weighting for class level comparison of the starts, that is to be expected.

As I recall, what you will find is that about the only meaningful lesson to be derived is that the world read Dr. Quirin's book and that 7-of-10 shows or good races, or 4+ winners in the last 10 races now produces a horrifyingly low return.

[raybo]

Ray,

I think you have my Percentages & Probabilities book. If so, look up the Earnings Box Class chapter.

If not, for now I will simply say that "consistency" produces highly disappointing results.

Absent some kind of weighting for class level comparison of the starts, that is to be expected.

As I recall, what you will find is that about the only meaningful lesson to be derived is that the world read Dr. Quirin's book and that 7-of-10 shows or good races, or 4+ winners in the last 10 races now produces a horrifyingly low return.

Yeah, that makes sense, but I'm not really looking for consistency, per se, rather relationships, if any, that enhance existing profitability. And, not necessarily win betting.

[Cratos]

Yeah, that makes sense, but I'm not really looking for consistency, per se, rather relationships, if any, that enhance existing profitability. And, not necessarily win betting.
To me this is a nonlinear 3-D problem between winning mutuel odds (winning includes place and show payoffs), frequency(probability) of the horse winning, and ROI.