Lest the gentle reader think that the great statistical curmudgeon cleaves only to numerical purity, he also conjured a method to estimate quality of a racehorse’s performance with performance points.
Sidebar Two — Performance Points
In order to ascertain how good a stakes winner (or group of stakes winners) is, another statistic was invented. Each stakes winner is awarded 400 points for each G1 victory, 300 for G2, 200 for G3, and 100 for all other nongraded stakes wins (all by black-type rules). They are also awarded an additional point for each $1,000 earned.
Take Pleasantly Perfect, for example. At 9,890, he is best stakes winner among this group. He won nine of 18 starts and earned $7,789,880. Six of those nine wins were in stakes, three in G1 and three in G2. That gives him 2,100 points, plus 7,790 points for his earnings for a total of 9,890. This statistic is called Performance Points, and the group statistic derived therefrom is called the Performance Points Index (or PPI for short).
The main problem with using earnings is inflated Japanese earnings. Japanese earnings have been divided by 8.5 to bring them back down to reality. For a more detailed discussion of Performance Points and the decision to deflate Japanese earnings by 8.5, see my series on the Rasmussen Factor on this website last August.
The Performance Points Index (PPI) is the average number of Performance Points per foal for each group divided by the overall average number of Performance Points for the entire group.
Let us use some hypothetical numbers. Say the entire group is 50,000 foals, of which 2,000 are stakes winners, and those 2,000 stakes winners earned a total of 1,200,000 Performance Points, or an average of 600 each. A subgroup of the entire group consists of 5,000 foals, 250 of which were stakes winners, and those 250 stakes winners earned a total of 125,000 Performance Points, or an average of 500 each.
So the subgroup had a higher percentage of stakes winners from foals than the overall group (5% to 4%), but their stakes winners were not as good as the overall group (500 to 600). Taking both factors (quantity and quality of stakes winners) into account, I say that since the subgroup is 10% of the entire group, it should have 10% of the 1,200,000 Performance Points of the entire group. That would be 120,000. The subgroup actually has 125,000 Performance Points. So I divide 125,000 by 120,000 and say that the subgroup has a PPI of 1.04, meaning that it is slightly better than the overall group.