You know about the “smell test”—something either rings true, or it doesn’t. In other words, it corresponds to our perceptions of reality. In the world of Thoroughbred breeding sire analysis, this is the problem with the average-earnings index: Even though the definition and structure of it are mathematically correct, the results don’t really ring true. I guess not too many people notice it because the average-earnings index isn’t used very much these days except in ads, but, in my small world, it was never really useful. I would see sires with a 1.40 average-earnings index, but I knew they were only average sires. So, for me, the average-earnings index, theoretically pure as it might be, never passed the smell test.
In working on developing the APEX method of rating sires (denoting Annual Progeny Earnings indeX) at Racing Update in the 1990s, we made an interesting discovery: By knocking out sires who didn’t average at least 10 foals per crop (that was roughly 20% of the foal crop at that time), we ended up with averages which, when we looked at them, 1.00 = average. As things evolved, sires were included only if they had 10-plus 3-year-olds of the last year covered. You can argue it’s less theoretically pure, but whether or not that’s the case, the results, to me, pass the smell test because they correspond with what I perceive reality to be.