A quick follow-up on ‘What have we learned from Free Agency?

I got this question from a reader and was interested in the answer myself, so I looked into it.

538’s defensive measure, aptly named DRAYMOND, was developed as a better way to measure defense and may give us some insight into how the league values defense. As I covered in the original article, average salary in free agency had essentially no positive or negative correlation with defensive performance by Jacob Goldstein’s D-PIPM.

Using 538’s DRAYMOND, the correlation improves significantly — average salary and DRAYMOND are correlated at a coefficient of 0.253 across 61 free agents who qualified for DRAYMOND (minimum 10,000 possessions in the last five years). That’s a lot better than the -0.021 correlation between D-PIPM and average salary.

Cutting down the D-PIPM list to just those same 61 players reveals no more information about D-PIPM’s correlation with average salary; the coefficient there is -0.016.

These results don’t necessarily mean that DRAYMOND is a better defensive metric than D-PIPM. Rather, it means that DRAYMOND lines up better with how the league seems to value defense on the free agent market, though there are obviously issues with sample size across 61 free agents in one year’s market. It’s still not clear how the league is evaluating defense in their free agency decisions; it may be entirely random that DRAYMOND correlates better with this market than D-PIPM does. As we go through this exercise over the next several years, we’ll have a much better idea of which metric is “better” with respect to the free agency market.

The full table of DRAYMOND tiers (broken up by half standard deviations, like I did with D-PIPM) is below:

Tier Years Total Salary Average Salary
Superstar 36    815,095,363      22,641,538
Elite 20    495,690,307      24,784,515
Above Average 21    237,194,700      11,294,986
Below Average 46    599,353,753      13,029,429
Bad 15    296,266,600      19,751,107
Horrible 32    536,229,222      16,757,163