Here’s my update on the polytrack surface at Keeneland. 

 I recorded results for the last five days at Keeneland and Belmont.  I measured how many times my top speed score (as determined by e-ponies.com) won a race and how many times my top speed score outran his ordinal position.  Outrunning the ordinal means that if my top speed score was the fourth choice in the morning line, he would have to have run better than fourth to outrun his ordinal position. 

Here is the data in tabled format:

                 Races       Wins      Outran Ordinal
Nonpoly      45             12           17
Poly            47              4            16

This data would indicate that maybe Mr. Beyer was correct.  But to be sure, I ran a simple hypothesis test on the data.  Here are the results:

Test and Confidence Interval for Two Proportions

Sample      X      N  Sample p
1          12     45  0.266667
2           4     47  0.085106

Estimate for p(1) - p(2):  0.181560
95% CI for p(1) - p(2):  (0.0297125, 0.333408)
Test for p(1) - p(2) = 0 (vs not = 0):  Z = 2.34  P-Value = 0.019

These results indicate clearly that the number of winners from my top speed score is lower at Keeneland over the test period. 

But if Mr. Beyer’s suggestion is valid, speed horses should be handicapped from running well, as well as from winning.  So I performed the same test on the data which compared my top speed score outrunning his ordinal position.  Here are the results:

Test and Confidence Interval for Two Proportions

Sample      X      N  Sample p
1          17     45  0.377778
2          16     47  0.340426

Estimate for p(1) - p(2):  0.0373522
95% CI for p(1) - p(2):  (-0.158654, 0.233358)
Test for p(1) - p(2) = 0 (vs not = 0):  Z = 0.37  P-Value = 0.709

These results are also very clear.  They indicate that horses are not any more or less likely to outrun their ordinal position over the polytrack surface as they are over the Belmont surface.  In other words, there is no difference.

So that is the statistical conclusion.  What are the practical conclusions of this quick study?

Don’t make snap judgments about the racing surface.  While the first test shows that winners from my top speed score are lower at Keeneland compared to Belmont, there could be a variety of factors as to why the horse did not win.  The top speed horse may still have run very well.  The second test is much more indicative of the impact on speed overall.  And it definitely says that the tracks are not necessarily and different in how it favors speed.  So don’t go changing your handicapping style over a Polytrack surface. 

I will run similar tests on any polytrack surface in use.  Once we have about 300 data points for each surface the results will become very reliable.