I looked at all RBs drafted in the first three rounds from '95 to '07, that at least took a snap in the NFL (Maurice Clarett and Kenny Irons> sorry). I counted the rookie year for the player as their first year where they took a snap (Ki-Jana Carter, McGahee, Foster, Shelton, Droughns, Bryson and Chapman were thus considered rookies a year after their peers in the draft).
...
Okay, my time is limited, so more data and less explanation. I looked at 7 variables (I'll explain why later) of production: Adjusted Yards over 3.0, Rushing yards/game, total yards per game, FP/G, games, games started and rushes. I looked at two pairs of input variables -- draft position and weight, and draft position and BMI.
Draft Position and Weight:
AY3.0: R2 of 0.13; weight variable is -0.70 (meaning you'd predict .7 fewer AY3.0 for every pound of weight); that's not large at all, and it's not even close to statistically significant.
Rush Y/G: R2 of 0.22; weight variable is -0.04; P value not close to significant
Tot Y/G: R2 of 0.24; weight variable is -0.16; P value not significant.
FP/G: R2 of 0.21; weight variable is -0.01; not significant.
G: R2 of 0.05; weight variable is -0.03; not significant.
GS: R2 of 0.19; weight variable is 0.01; not significant.
Rsh: R2 of 0.23; weight variable is -0.21; p value not significant.
Quick summary: Weight doesn't look at all relevant to rookie success, at least once you factor in draft position.
BMI and Draft value again:
AY3.0: R2 of 0.15; BMI -17.1; P value is significant at the 5% level. This is pretty interesting.
RYPG: R2 of 0.22; BMI is -1.28; P value not significant.
TYPG: R2 of 0.25; BMI is -2.52; P value significant at the 10% level.
FP/G: R2 of 0.22; BMI is -0.313; P value - 17% (not significant, but close)
G: R2 of 0.04; BMI is -0.17; not significant.
GS: R2 of 0.19; BMI is -0.05; not significant.
Rushes: R2 of 0.24; BMI is -4.45; not significant.
Couple of thoughts:
1) This only looks at rookies. And all rookies. This is distinct from the question of whether backs that are good eventually start out differently based on BMI.
2) To the extent that the "goodness" of BMI is nonlinear, this study doesn't answer that question. Some contain, and they're probably right, that there's a sweet spot of BMI between about 28 and 32, IIRC. So a 34 BMI and a 26 BMI are both "bad". This can be examined when I get home in a few hours, along with Q1 (and a full explanation of what half this post actually says).
Let me actually explain that post now.So I looked at highly drafted rookies the past decade or so, and saw how they did their rookie seasons. Obviously higher draft picks are going to outperform lower draft picks, on average, but the question is do bigger backs outperform smaller backs early on, after accounting for draft position. To do this, we need to define "bigger" and "outperform", since neither are clear.
We can perform a regression analysis to isolate "big" from draft position when judging performance. I think "big" can be defined using either weight or Body Mass Index, which measures weight as a function of height. Let's use weight first.
There are 7 different ways I looked at measuring performance. Games, games started, and rushes are good raw measures of performance and value, but they don't measure ability. Rushing yards per game, fantasy points per game, and total yards per game are really good measures of ability. My favorite stat mixes both volume (like games or rushes) and performance (like FP/G) -- adjusted yards over 3.0. It's simply calculated by taking a player's rushing yards, adding 10 yards for every rushing TD, and subtracting three yards for every carry. So a player that rushes for 1,000 yards and 10 TDs on 300 carries will have 200 adjusted yards over 3.0, just like a player with 500 rushing yards and 0 TDs on 100 carries. We could argue all day which player is more valuable or more talented, but I think it can be reasonably argued that their equivalent. Anyway, those are the seven stats I used.
Looking at weight and those seven factors, absolutely none of them were correlated with weight. So basically, a 230 lb RB drafted with the 19th pick shouldn't be projected to play any differently as a rookie, than a 210 lb RB drafted with the 19th pick. The closest variable to being statistically significant was Games and Weight, where maybe a player for every extra 20 pounds can be expected to miss half a game.
It's not surprising that the results are all insignificant, because weight isn't a great proxy for "bigness". I think BMI is a much better one.
BMI
is statistically significantly correlated with rookie production, although all of my BMI comments have a big caveat discussed at the end. But for every additional one point in BMI, you can expect a RB to have about 17 fewer adjusted rushing yards over 3.0. That's not practically worth much, but between someone like Jamal Lewis (BMI of 33.5) and Reggie Bush (BMI of 27.1) we might expect (assuming they were the same draft position) about 109 more adjusted yards for Bush. And that's not significant, although it's not incredibly valuable, either.
BMI wasn't correlated with rushing yards per game, although it is arguably correlated with total yards per game. This may simply be because smaller, thinner, faster RBs are better receivers than big RBs, and I'm not sure if this has anything to do with being a rookie or not and smaller RBs adjusting to the game quicker.
FP/G and BMI may be correlated; the results weren't statistically significant, but they were close. Using Bush/Lewis again, we might expect Bush to average 2 more FP/G than Lewis, assuming the same draft position. There's a decent change that's just noise in the data, but it does make one thing clear: it's unlikely that the heavy RBs are better early on than light RBs. They're probably a push, or thin RBs are slightly better.
G, GS and rushes weren't connected with BMI.
So in conclusion, it looks like the ability stats as opposed to the compiling stats, are connected with size. And to the extent that there's a connection with size and rookie RBs, lighter is better. The thing with BMI, though, is that there's probably a sweet spot in the middle. I was rushing earlier, but I should have accounted for that. I'll do that soon.
