How much can the NBA draft combine measurements be used to “predict” defensive success in the NBA? I’m enough of a nerd to try to figure out.
I ran some correlations with Basketball Reference’s NBA career defensive ratings against the Draft Express’s Measurement History database and found the following:
PG: wingspan and vertical reach (but the total wingspan and reach gave only minor effects; it was the ratio between wing & reach to height [non-alligator arms] that was more important)
SG: wingspan, reach, agility test
SF: weight (more is better); vertical reach
C: wingspan, vertical reach, jumping vertical reach with no step, and maximum jumping reach (but interestingly the ratio of wing & these various reaches to height was more important, especially for the jumps, than total wing & reaches); bench press (but more presses meant a worse defender)
PG: weight (less was better); no-step jumping vertical reach; maximum jumping reach; bench press (fewer was better); sprint (slower was better)
SG: height; weight (more was better); no-step jumping vertical reach; maximum jumping reach; sprint (slower better)
SF: height; no-step jumping vertical reach; maximum jumping reach; bench press (fewer better); agility (slower better)
PF: wingspan; vertical reach; no-step jumping vertical reach; maximum jumping reach
C: weight (more better); wingspan and reach (though ratio of these to height was as or more important than total wingspan & reach)
1) Measurement/testing results had greatest correlation with SF defensive prowess, least with PF
2) Jumping ability is overrated. Verticals (not including reaches) had no impact on defense at any position. Jumping reaches were less important than standing reach, except for centers
3) Ratios of reaches and wingspan to height was at least equally important as total jumping reaches and wingspan in some instances
4) Bench press is a useless test for predicting defense
5) Agility test only seems to matter for SG
6) Sprint matters very little
7) Height has only very small impact; weight is a better predictor of defensive success (though for PGs less is better)
Thoughts about the results? Do they alter the way you look at any of the 2013 prospects?
Some even more nerdy details:
1) Used the entire DX Measurement history database of players that received more than about 200 NBA minutes. About 375 total players, approximately 75 per position.
2) Many potential problems with this research, including: sample is problematic since many NBA players (esp. the best and the worst) didn’t go through the full NBA combine experience; defensive rating a problematic stat; taking career defensive rating perhaps is perhaps biased against both players who are now young and those who had long careers. But I think the sample size is probably large enough to wash away many of the issues.
3) Correlation coefficient of more than 0.3 meant major effect for me; between 0.3 and 0.2 meant moderate effect; between 0.2 and 0.1 meant minor effect. None of the correlations, except for the SF wingspan (0.412), can be considered very large.
4) Correlation coefficients are not very helpful in seeing relationships to defense that may be non-linear (such as for weight or bench press, where it is better to be in the middle than on either extreme perhaps).