Posted by kmasonbx on 7/1/2012 10:46:00 AM (view original):
Yea, I remember when LM said that he thought CVs were 2.5-3x HVs, but I'm pretty sure he was wrong. I think the problem is back then, and even now, there is a misconception about the advantage of prestige. I remember it was going around that in D1, 1/3 letter grade was a 33% advantage, and I think now most people realize that that's far from the truth. People even thought there was a decent advantage at D2/D3 but now most people realize that when it comes to battles prestige is nearly meaningless outside of D1.
i agree with this too. i have a lot of respect for lostmyth, as hes a great coach, arguably the best ever, and easily hold the argument of best d1 coach ever, if there is even a distinction to be made (i would argue there is, but i can see the other side).
however, you have to look at the WHOLE of his responses on the subject.
1) prestige is worth 70% per letter grade
2) prestige scales linearly
3) HV:CV is what, 2.5-3?
4) prestige is not capped on d1 a+
you have to take what lostmyth says in light of his situation.
1) he basically only coached with an a+ for a long while
2) he claimed prestige *must* scale linearly due to the nonsensical battles he would have to explain based on smaller contrasts in bigger prestige battles
well, the linear prestige claim has been proven false. there is no logical system where you can have a linear prestige progress. as in, if A to B = 2x, A to C = 4x. or else what in the hell would you do when an A, B, and C schools battle? there is a simple conversion to effort, virtually everyone understands that. so if the B = 2C and A = 2B, then A = 4C but also in a linear system, one would claim A = 3C. contradiction.
so, knowing now that lostmyth had a flaw in his understanding of d1 prestige, whats the impact on the rest of his claims?
so, heres what i claim. any battles within a single prestige grade aren't worth half a ****. (note, there are exceptions, but they are rare and most coaches cannot spot them). figure A to B, is that 1 prestige grade? or is the A a low A, the B a high B, making the difference of a low B+ to high A-, which is 2/3rds. or is the A a high A, and B low B, which is 1 1/3 grades - twice the low end range. so with a factor of TWO variance on the damn prestige grade, which is a major factor, it is *incredibly* easy to have misleading results.
anyway, one thing is obvious to me. if you think prestige scales linearly at the high range of battles, you were over valuing prestige. multiplicative prestige scales very quickly. if, on a wide prestige battle lostmyth was going hmm this range of prestige wasn't as big as i though, so it must scale linearly - it stands to reason, his value of prestige is probably high.
well, here is where i don't have a tidy logical reason to break down his HV:CV ratio, which as you might have guessed, i disagree with (i only agree with his claim about a+ not being capped - goes to show - this **** is not simple, if you can, at one time, take guys like OR, LM, and myself, and we can all heartily disagree with each other on prestige range). one would think a high value on prestige would lead to an under valuing of whichever tool one used - and CV seem to be LMs tool of choice (or maybe, it WAS HV, but his value on prestige drove him to believe CV > HV at low distances??). so, all i can say is this:
this **** is NOT simple. im a math guy, a computer algorithms guy, and in this world of basketball simulations, THAT is my strength. guys like OR crush me on, you know, having a damn clue how to play basketball (i have none). i can say honestly, i was surprised how difficult it was to attempt to nail down prestige. the range of variation is so huge, its so damn easy to get sucked in to a frame of mind but when you have things like a wide prestige range (a whole grade, about the largest you see with any regularity in d1 - STILL being a factor of 2 variation) - its insanely difficult to pin down! but its easy to go, i think prestige is 1.7, HV:CV is 2.7, and most battles will fit those values - because of the massive ranges involved.
so anyway, i would guess lostmyth has had some of the best d1 A+s of all time, and more of them than anyone, by a lot. any time you have an advantage like that, that you can't really count on, it can easily hide other losses - such as, using CV at 250 miles, when you should use HV, but experience only a moderate (about 20%) loss - his a+ could have caused him to win battles, that he attributed to CVs being stronger than they are. also, often, coaches make judgement calls based on what works. LM most likely was spending his money more efficiently than just about any coach, so when he would presumably, like all of us, "predict" if he could win a battle - he probably only ended up fighting battles he expected to win, without factoring in his efficiency (i do the same thing, im sure most of the other top coaches do too), and then he won. so why question his values? no reason, as best i can tell, tough to doubt yourself when you as successful as he was.
so i would just guess he fell into one of the many traps and pitfalls any coach can fall into trying to approximate prestige and HV:CV. *if you don't save your battles in writing you basically don't have an ice cube's chance in hell, due to the extreme uncertainty involved*. did he save his? i dont know, but its really easy not to. your *ONLY* chance is to take battle after battle, compute the extremely wide ranges of possibilities, always considering the possibility the other coach, who most likely just lost, is ****** off and lying to you (or made a mistake). when you get enough of these extremely wide ranges, they do prune things down, and over time, you can narrow in. BUT if you have value in mind, that is reasonably close to the real values, you usually can find a way to make it fit, its just what happens when there is such wide variation. it just simply takes a mathematical rigor to really nail this down that most people are not used to. when you do get a tight battle, that maybe shows you were off (say it was 1.6 not 1.7, might take 20 battles to find out you are wrong) - its close enough, and you've had so many more positive indicators, its easy to wave one slightly negative one off.
for some semblance of completion, i should also mention - you also have the possibility of finding these things other ways, other than battles. but i doubt more than a couple coaches have figured out those methods. surely, they exist, but again - with an incredible array of possible pit falls and challenges. and a similar level of mathematical rigor is required, along with a detailed understanding of the mechanics of recruiting. and its just the reality of life - most of the "sports people" who tend to excel quickly in this game, with a strong intuition for sports - are not the math geeks who take the time to prove everything under the sun. so i just dont think its very likely that many coaches would figure out how to *really* nail down things like prestige value, and HV:CV.
i mean, to illustrate my point, i can only think of 1 really good example. OR, many times, posted that he thought prestige was 2:1. and he now has retracted that, saying he doesn't think it can be close to that big. there are few coaches, if any, who have a better understanding of this game than OR. hes coached basketball and has done technical work, and of course, has a dozen more championships than any other coach. so if someone like him can make the mistake - ANYONE can. the only way to figure this out is through incredible rigor and open mindedness - and the reality is, most of the best coaches are those with the best intuition. and those with great intuition lean on it, hard (as they should). but things like prestige, HV:CV, are just incredibly conducive to making mistakes, if you are leaning on your intuition. so its very easy to do. and i dont think most coaches honestly care enough to actually go figure out what the HV:CV ratio is, nor what the value of prestige is, and it that case - you may as well take the opinions of a few top coaches, and pick one, or average them, or something. but if you *REALLY* want to know, you just can't take the word of anyone, doesn't matter who it is.