Quote:
Originally Posted by dilanesp
He has a very mixed legacy.
He certainly cared about the game, wrote about it, and studied it. And he was very intelligent.
But Dosage was a classic example of the over-statisticalization of the game based on overly small samples. And every year, around TC or BC time, I hear a bunch of BS statistics based on small samples and retroactive fitting (such as that Exaggerator couldn't win the Preakness because Derby runners-up never do). And Dosage was the grand-daddy of such theories-- a completely arbitrary, back-fitted theory that tried to predict something that had a tiny sample size.
|
I disagree in part with your criticism of Roman’s methodology.
At its inception Roman’s methodology made statistical sense and was predictable when sires were limited to covering 32-36 mares per breeding season and the big farms dominated the TC races by breeding prized stallions to their very good mares.
This is not to say that the breeding today is inferior to yesteryear’s breeding, but it is to say that it is different and doesn’t fit Roman’s Dosage model.
As you very well might know that statistics is not a science but an evolving data driven methodology underlined by many of the other sciences.
Therefore many concepts put forth in racing today for handicapping and breeding is inept because of statistical evolution and modernization or lack of supporting science.