This is what I get for average beaten lengths at the finish.
Code:
Surf Dist Races Avg BL
1-Dirt 1-Spt 106333 2.55
1-Dirt 2-Rte 43351 2.89
1-Dirt 3-Mar 61 2.73
2-Turf 1-Spt 9132 1.54
2-Turf 2-Rte 15829 1.49
2-Turf 3-Mar 191 1.50
3-Poly 1-Spt 10752 2.05
3-Poly 2-Rte 6191 2.21
Sprint is < 1 mile, Route is
> 1 mile and
< 1 1/4 miles, Marathon (LOL) is > 1 1/4 miles.
This definitely argues for the popular beaten length adjustments changing too much as distance changes as steveb suggested.
What is interesting is that the value of time seems to need two measures IMO. You need one to rate horses behind the winner, and another to rate the winners and be able to compare races run at different distances.
You could easily assign a value of one point to, for example, 0.15 seconds and apply that for all distances on dirt as a "beaten lengths" value. A horse beaten 0.60 seconds is rated 4 points below the winner.
What you can't do is use that same 0.15 seconds and apply it to winning times. If you have calculated a baseline of 1:10 for 6f and 1:22 for 7f, both equal to 100, you'll get in trouble if you try to use that same 0.15 at both distances for the winners. For example, if a 6f race goes in 1:12.50 and a 7f race goes in 1:23.50, both are 1.50 slower than the baseline. If you use that same 0.15 both races would be docked 10 points and be rated as 90, but in reality the 7f races should be higher rated, all other things being equal.
