How should someone approach the subject of speed in a race that is dominated by stamina? That's a question I've struggled with for weeks as I prepared a statistical profile for the Belmont stakes. The vast majority of my normal speed measurements don't seem to apply.
It has been a basic tenet of my statistical models to start with averaging the winning performances of all the entrants in similar and applicable races, then requiring that each horse equal or better that mark at least once in their lifetime. This approach simply does not work with the Belmont. If one were to take all the winning stakes route performances by Belmont contestants into account then apply them against the figures earned by the entire field you find that many winners were "too slow" to compete. 39 of the 85 Belmont contestants from 2000 to the present failed to achieve the field average route stakes winning Beyer Speed Figure. But those 39 horses combined for 4 wins and another 5 in the money placings. All the best longshots like Sarava, Da' Tara and Birdstone would have come up too slow.
My supposition is that the measurement does not work for this race because we don't have any relevant data. 12f is a long race and figures earned even in 10f races like the Derby may not be that applicable since we're talking about adding another quarter mile. I know that when looking at 9f races I'm not inclined to use Beyer Figures earned at 7f because the distances are too different to be comparable. Virtually no Belmont entrant has ever run more than 10f prior to the race and an extra quarter mile can produce a major difference in the level they are able to achieve.
Most horses run slower at 12f than they do at shorter distances so figures produced at shorter distances must carry less weight. The Belmont has failed to produce the highest winning Beyer Speed Figure among the three Triple Crown races in every year recorded aside from Lemon Drop Kid in 1999 and A.P. Indy in 1992. That means the Derby and Preakness are almost always faster races.
One can actually make money by simply betting the horse that earned the fields highest Beyer Speed Figure in their last race in the Derby and Preakness. Not so with the Belmont, you'd be down 50% from the year 2000 to now.
Some people have written about the fact that closers don't do well in the Belmont. This is a fact pulled out of thin air. The writers probably just wanted to sound interesting but it has no statistical relevance. For every front runner like Da' Tara and Commendable there is a closer like Jazil, Sarava or Victory Gallop. There is no statistical bias for or against tactical speed in the Belmont
So does speed have any place in the Belmont? I have to say that Form and Fitness are much bigger factors in my mind and speed (figures and tactical) are virtual non-factors. If there is any race where it would be wise to ignore the bolded figures in the center column of the past performances this would be it.