Dosage Index Stuff

So this is pretty theoretical and technical stuff but is something that has intrigued me for quite some time now. When you are looking to breed a mare you almost always do what is called an ENick to figure out what the theoretical cross is rated as. I did Rachel Alexandra and Any Given Saturday just as an example. It spits out this:

Roman Dosage Profile: (2-3-6-1-0) (distance from Sprint-Classic-Marathon, basically)
Dosage Points: 12
D.I.: 2.00 C.D.: 0.50
Werk Quality Rating: Q 5
Quality Points: 10 Total Points: 22

So what the hell does this all mean? There is a great explanation of all of it here. I don't really want to get into all of it but basically most of these numbers are just based on the occurrences of the chefs-du-race, or the chiefs of the breed (you can see who they are at the bottom of the link). But what about all of the other blood in the pedigree? (And really, who cares about these numbers, you don't get more money at the sale solely because of them).

So yeah, it's awesome that Mr. Prospector, or Norther Dancer, or Seattle Slew show up 10 times in a pedigree, but what about the other 116 horses that make up the previous 6 generations of any given horse? Are they just discounted? Or does the bloodline only count if a recognizable name shows up? How is Animal Kingdom rated with the "small" names that litter his bloodline? I've always thought there was a better way to actually quantify the "quality" of the blood in any give horse by using the basic numbers that are out there.

Now there are great programs out there to really do the leg work insofar as looking at many generations of possible crosses. TesioPower is the one I've messed with but I don't have access to it currently. But it *should* be able to do this if you put it together properly.

So I think the baseline value should be built on the methodology of Col. Jean-Joseph Vuillier, who was really the first guy to try to really quantify the influences of specific bloodlines in thoroughbred racing. He went back 12 generations, but I think in our day and age you'd probably reach the common ancestor by about 6-8 generations based on all of the inbreeding that has occurred. So let's do 6 for the sake of argument.

Just like Vuillier I'll assign 1 point to the 6thgen, 2 to the fifth, 3 to the fourth, 4 to the third, 5 to the grand-sire and grand-dam, and 6 to the sire and dam. Now that's great and all, but we need to multiply their "point(s)" by how productive and successful they were/are bothon the track and in breeding. So here's my proposal:

Weight = Earnings/10,000,000 + AverageProgeny Earnings/100,000 + (Stakes Winners per foal*2)

*Note: All Earnings would be inflation adjusted for 2010 dollars

So some multipliers:

• Mr. Prospector: 544213/10000000 + 155607/100000 + [181/1099] *2= 1.940
• Roman Ruler: 1355088/10000000 + 23572/100000 + [7/257]*2 = 0.426
• Rowdy Angel**: 16395/10000000 + 628250/100000 + [2/8]*2 = 7.084 (definitely too high)

**Rowdy Angel produced Demons Begone and Pine Bluff, who were both stakes winners and earned 609k and 2.25mil (in unadjusted dollars) respectively, so her numbers are really, really high based off of her two huge hits.

Now all of that took me a while to figure out since I had to do it by hand. I just used a large number as a denominator in order to put everyone on an even playing field. Earnings are 10mil since they're less important than progeny earnings, which is 100k. Stakes Winners/foal are multiplied by 2 since that's really what you're aiming for. I really wanted to use median earnings for the progeny number, but there are just too many unraced horses out there for that to really be a meaningful number. You can't just discard non-runners since that speaks not just to durability and disposition of foals, but also to how valuable the bloodline is since so many horses (especially mares)  immediately go to be bred.

Once the "weight" is figured out you multiply it by the "points" and add it all up. The more points, the better the bloodline in the foal you're trying to create in a hypothetical mating. This obviously is not perfect, as no mathematical computation possibly could be when looking at this, else people smarter than me would have already figured it out.

You can also add in a multiplier for 2yo starters/earnings etc. It clearly discounts the younger sires due to small sample size. But new sires are always a gamble anyways.

Now the entire pedigree's weights will inherently take into account the success of the offspring of that immediate ancestor while ignoring the siblings that are on the same crosses (So MP factors in his immediate winners, but not those of FuPeg or Woodman). My biggest issue with the dosages and profiles et al is that they discount the "other" horses that make up the vast majority of a runner's pedigree. Look at Rowdy Angel: She didn't win much, was well bred, but certainly not impeccably so, but she produced two stakes winners and nearly 5+mil in inflation adjusted winnings. That's impressive but won't register anywhere in the current system.

The big hold back is that you would have to do these calculations on every single horse that ever ran in order to be able to paint an effective and usable picture, so it might be too much to actually handle.

This may also weight mares too highly if they have one heavy hitter, but that's dealing with small sample size per mare no matter what you do. So maybe you could just do this with stallions, and add in points for really exceptional mares (like Rags, Zenyatta, Rachel, etc). You could use the Reines Du Course listing and just double their points(so the grand dam gets 10 instead of 5 as a reines du course) when adding it all up.

This probably interests no one, and there are a million holes in it, but it's just something I thought needed some explanation, at least in my head. But this is definitely still a work in progresss