Playing Around With Prime Power Numbers

[classhandicapper]

Originally Posted by Robert Goren

Plus I have some doubts about that a point difference between the top two horses is the same as point difference between the bottom two horses.

That's a pretty interesting insight.

At the top, you are more likely to encounter horses that were suited to the pace, got a reasonable trip, and were ridden out to the wire.

At the bottom, you are more likely to have issues with time/beaten length formulas, horses that were eased, horses that encountered horrible trips or that were used hard in the pace, horses that were overmatched on class etc.... So it's tougher to know who was actually better even if they came out of the same race, let alone different ones.

[Robert Goren]

Originally Posted by raybo

Why 0.14863? Why not use the PP as it is?

That is the number I got when I did logit regression. For each point difference in prime power the chances. the chances of that horse winning goes up by about 0.157%
if you did not multiply the PP by 0.146, you would get very different Odds. Somebody might get a different multiplier. There are a lot of short fields in my sample. I am current looking to see if the multiplier changes the futher you get from the mean PP (or the top PP number) in the race. I am pretty sure it does. Eventually I will get around to looking at how the tote board odds effect a horse chances of winning at different PP places in a race. Does the tote odds effect the chances of an 8/5 shot (as figured by the PP) differently than it does a 9/1 shot? That is question I want an answer to eventually. Thanks for your interest.
PS I tried to insert a table from excel to show what I did , but I am not smart enough to do it.

[Robert Goren]

Originally Posted by classhandicapper

That's a pretty interesting insight.

At the top, you are more likely to encounter horses that were suited to the pace, got a reasonable trip, and were ridden out to the wire.

At the bottom, you are more likely to have issues with time/beaten length formulas, horses that were eased, horses that encountered horrible trips or that were used hard in the pace, horses that were overmatched on class etc.... So it's tougher to know who was actually better even if they came out of the same race, let alone different ones.

I suspect that if you get far enough away from the top PP in a race, that the point different doesn't matter. Somewhere out there about 30 points from the top horse a point difference does not effect the chances of winning because all the horses have the same chance, a very small %. PP can not predict a black swan event or at least I doubt that they can.

[classhandicapper]

Originally Posted by Robert Goren

I suspect that if you get far enough away from the top PP in a race, that the point different doesn't matter. Somewhere out there about 30 points from the top horse a point difference does not effect the chances of winning because all the horses have the same chance, a very small %. PP can not predict a black swan event or at least I doubt that they can.

That's actually another insight.

I was talking about the horse's previous races being used as input today. It's harder to trust figures on horses where they were well beaten.

But you are probably correct again. There's probably not much of a difference in the chances of winning between horses way at the bottom of the field because it's going to take a huge wake up to win and they probably all have roughly the same chance of waking up that much once they are rated poorly enough.