A few weeks ago I wrote about a fitness angle I use called Competitive Race Last Time. A couple of small tests I did seem to indicate that horses tend to regress or bounce after running a race where they were within a couple of lengths of the leader at the stretch and/or finish. Another fitness angle I use, Last Race Equal or Greater than Lowest Bounce Point, also appears to point toward horses who are likely to bounce today.

Gary Voekl pointed out in The Grandstand message board that it was impossible to know what caused this bounce, and that it might be caused by a class rise after the good race. It could be that the horse tends to be placed in too tough a spot after a good race, and that�s what causes the bounce, not physical stress and fatigue. It was a good point, so I decided to do a study on this subject. I looked at another sample of 114 performances where a horse ran a race equal to or greater than their "lowest bounce point."

(To quickly recap the Last Race Equal or Greater than Lowest Bounce Point fitness angle, the logic behind this one is that there is a certain level of performance which will cause a horse to regress or "bounce" next time. Say you have a horse with this set of Beyer speed figures in its past performances (in chronological order): 67, 74, 76, 90, 85, 89, 61. What I'm looking for is the lowest figure where the horse always regressed the next race. The horse in this example improved over the first four races in its pps, then dropped from 90 to 85. He then improved to an 89, then regressed to a 61. Note that the highest figure this horse ever improved off of is the 85. The lowest bounce point has to be higher than that. The next highest number was 89. That's the "lowest bounce point." Whenever the horse ran an 89 or higher, he declined next time out.)

In this new sample, like all the others, horses tended to perform worse after a race equal to or greater than their lowest bounce point:

All
Improved/Same:  31/114     27%
Declined:             83/114     73%

I also broke the sample down according to whether the horse went up in class, stayed at the same class level, or went down in class in the next race:

Up in Class
Improved/Same:    13/49     27%
Declined:               36/49     73%

Same Class
Improved/Same:    10/41     24%
Declined:               31/41     76%

Down in Class
Improved/Same:     8/24      33%
Declined:              16/24      67%

As you can see, horses that went up in class the next race after the stressful race improved and declined at exactly the same rate as horses in general. Horses that stayed at the same class also improved and declined at pretty much the same rate. Horses that moved down in class actually declined a little less often, but the sample is too small to be definitive. There�s no evidence that class changes have any effect on bouncing. Horses tend to bounce after a stressful race and it seems to have nothing to do with class moves.

In a way, it really doesn�t matter what "causes" bounces. The important point is that it happens, and that despite most handicappers having heard of the term, the general public still isn�t really on to it in everyday situations. As Gary Voekl pointed out in another message, it may be regression to the mean (a tendency to return toward an average performance) that causes this. But regression to the mean doesn�t occur magically � it happens due to forces at play in the situation being studied, like physical stress acting on a horse after it runs a hard race.