How to rate stallions - Elite above Expectation
For some time, probably as long as I have been in this industry I have thought that the metrics that the industry uses to measure stallion success or genetic merit were pretty ordinary and not much had been done to improve that despite what has occurred in other sporting fields in terms of assessment metrics and in data science in general.
The Average Earnings Index was created by Blood Horse editor Joe Estes in the 1950's, and while more recently Bill Oppenheim developed the APEX ratings, neither of these ratings are without flaws and nothing has been developed to improve on them.
Both the AEI and APEX suffer from the non-normal distribution of prize money. Regional/State-bred racing where they are boosted by gaming supplementation sees stallions appear to be far better than they are (especially in the APEX rating). Equally one 'hyper earner' can make the AEI look silly.
I was inspired a bit by the World Cup and the concept of Expected Goals (xG). Its a little controversial as a metric, but I thought it was a sound way to properly measure 'opportunity' within a metric and thus show actually what stallions have genetic merit.
Click here to download my white-paper on it.