What is also interesting is that “A” nicks far outweigh any other category. This is perplexing, as it seems to me, that an “A” nick should be the nick that occurs least frequently. We certainly know that superior runners occur least frequently. It seems that by sheer numbers alone the “A” nicks are doomed to failure. What am I missing?
The above is a comment from Ned Williams. Here is my reply:
You are not missing anything. You are correct that A nicks being 40% of the entire sale (as opposed to 13%-15% of the overall population) sticks out like a sore thumb. And you are probably also correct that A was doomed to failure by its sheer numbers.
What people have a hard time grasping is that the Keeneland September sale is a pretty good sale overall. Even the pedigrees offered on the final day of the sale are not bad pedigrees (relative to the entire population, which is a comment on the lack of quality in the entire population). Sales foals are generally better than the entire population. The Keeneland September sale is better than most sales. Therefore, it is not really surprising to me that 40% of all nags sold in the 2010 Keeneland September sale were A nicks.
I have secured the links to the original cheatbooks for the 2010-2011-2012 OBS August yearling sales. That sale is much lower down on the scale than Keeneland September. Should not have nearly as many A nicks (although you never know). Might be a better population to study for that reason. A nicks might fare better in that environment. Will see. Worth a look.
Some other foolosophical reflections on the subject of nick ratings. As I understand it, TrueNicks tries to base their nick ratings on the sire-broodmare sire combination. If the computer finds enough foals to fit that description, they do base their ratings on that combination. Below is a quote from the TrueNicks explanation in the 2010 Keeneland September cheatbook:
“The TrueNicks program is designed to establish a rating within a minimum number of
generations from the sire/broodmare sire cross. As soon as a cross has met statistically significant thresholds, a rating is calculated. The farthest the TrueNicks program will go to calculate a rating is to the third generation of the horse (or prospective mating) in the sire line, and the fourth generation of the horse (or prospective mating) in the broodmare sire line. Ratings beyond those parameters would be meaningless and will yield a No Rating score.”
This written explanation does not specify what constitutes “statistically significant thresholds” (perhaps it varies), but I am guessing about ten foals. The only problem with that is that ten foals is really not a very big sample upon which to base a rating. And not many sire-broodmare sire combinations have 50 or 100 foals or more upon which to base a rating.
If the computer does not find enough foals, it goes back another generation or two (or more) to find enough foals and uses the total number of foals found for that particular combination. The only problem with this is that going back another generation or two (or more) makes any predictions based on that combination much more weakly correlated to the foal in question. If you have learned anything at all from this blog, I hope it is that “influence” recedes much more rapidly than most people think with each generation farther back you go.
So if you are in the business of making nick ratings scientifically (which not all of them do; some of them just skip the scientifically), you are in between a rock and a hard place. You can make a rating based on ten-plus foals, which may not be every reliable because that is really not enough foals. Or you can go back another generation or two and make a rating that way, which may not be very reliable because it is several generations removed from the foal in question.
In extreme cases you may not be able to make any rating at all because even going back several generations you can not find enough foals bred similarly to the foal in question. It must be pretty galling to discover that this not rated group actually had pretty good results (especially since buyers paid the least amount of money for them).
Here is something people may not understand about nick ratings. They are basically past performances. And as sellers of mutual funds are fond of saying, past performances are no GUARANTEE of future results.
Here is where I differ foolosophically from the makers of nick ratings. The latter might say, “This is an A+++ nick. It should improve your chances of getting a stakes winner by 100%.”
And I might reply, “Don’t think so. Maybe 10%-20%. Nowhere near 100%.”
So it is a question of degree. It makes perfectly logical sense to expect a mating (or a similar mating) that has worked well in the past to work well again. The question is HOW MUCH better results should you expect????? I think the makers of nick ratings exaggerate the amount of improvement they can deliver (not to mention its predictability). They exaggerate the importance of nicks.
I do not deny the existence of nicks. I merely think that their ability to deliver better results (not to mention the predictability of those results) has been exaggerated. Their importance has been exaggerated by people who are trying to make $$$$$$$$ doing so.
Those are some thoughts you might want to keep in mind as you go about your business of buying and/or selling at the Keeneland September yearling sale (or any sale, for that matter). Good luck at the sales!!!!!!!!!!!!!!!
That concludes this five-part series on nick ratings. Many of you seem to have missed the most important one, post number two. So here is a link back to that post if you want to use it.