Quick blog post and data drop re weight vs performance for WRs
Dataset = drafted WRs 2010-2018 (unfair using 2019-2021 data as they haven’t had much of an opportunity to get a Top12 or Top 24 season, and the S1-3 AVG PPG PPR data would also be unfair as had only played 0-2 seasons)
Weight taken is weight whilst on an nfl roster as some of the draft weights were missing and I need to be efficient with my time. Also, didn’t use BMI as the height data was in a format I couldn’t be bothered/smart enough to translate properly into inches.
Weight vs PPG PPR S1-3 (sorry, don’t have PPR PPG longer than S1-3, as that’s all we’re kinda interested in when scouting college WRs)
R^2 = 0.005 (Weight explains 0.5% of the variance of S1-3 PPR PPG). P-value 0.329 (e.g. 33% likelihood the findings are due to chance i.e. no correlation)
Weight vs Draft position
R^2 = 0.003 (Weight explains 0.3% of the variance of Draft Position). P-value 045 (e.g. 45% likelihood the findings are due to chance)
Edit 13/1/22: Re-ran analysis but with log of draft pick (should’ve done it initially, oversight)
R^2 0.015: Weight explains 1.5% of the variance of log draft pick. P = <0.05: This is significant! It means that there’s less than a 5% likelihood the findings are chance. So we can be confident that there IS a correlation between weight and log of draft pick, in that, as weight increases, draft position gets lower (or better) however the correlation is insanely tiny.
Divided weights into 4 quartiles based off the dataset. <191lbs, 191-202, 203-215, >215lbs, as people are usually quoting weight “thresholds” that they want players to meet. I used logistic regression for the next part (helps calculate an odds ratio).
Shouldn’t really have to account for draft capital as supposedly not any correlation between weight and draft position, but I did anyway as it’s the right thing to do.
Blue circled section is weight categories 2 through 4, compared to category 1 (<191 lbs). In order for it to be a significant difference, the blue circled values need to be less than 0.05. I think given the sizes of the samples we’re using though, I’m kinda on board with having anything less than 0.10 as significant. Basically, when accounting for draft round, none of the weight categories were more likely to record a Top 24 season than the other weight categories. Category 4 (>215 lbs) is the closest to significance compared to Category 1 (Odds ratio 2.05 but 95% CI 0.6-7), so maybe there’s a trend, but who knows what we’d find with a larger dataset, could solidify significance, or it could go back the other direction.
Edit 13/1/22: Again, re ran analysis but with log of Draft Position for Top 12 and Top 24 seasons – remains nil significant odds ratios
The same was found for Top 12 seasons, however even further away from significance.
Basically, I’m still rolling with I don’t think weight really matters enough to significantly influence my opinions on a player. Except maybe Tutu, come on man, I love you, but eat a cheese burger
Alb's Thoughts 