Just how do those of us on the outside evaluate how well a coach is doing? Even beat reporters in the Before Times didn’t have unfettered access to practices, film sessions, and meetings. We’re all limited mostly to what happens on the court during games.
Those limitations don’t stop most of us from asserting how stupid a coach is when the team loses. Obviously he should have just played the guy on the bench or a different lineup combination. Obviously he stayed too long with the bench. Or didn’t stay long enough. Or didn’t feed The Hot Hand. Or...you get the point. We all do it, including me.
Aside from obvious things like the team’s record or scoring differential, I want to experiment with some data-driven approaches to looking at NBA coaches in general, and Brooks in particular. With the Wizards on pandemic-induced hiatus, I’m diving into a multi-part series on evaluating coaches. Don’t ask me how many parts — at least one more after this.
This mini-project will be a work in progress that I’ll update periodically. I’ll talk to coaches and league contacts at some point. I’ll also revise the methodology and try new and different approaches as the season progresses.
Enough preamble. Let’s get started...
One of the biggest tools a coach has to influence his team’s success is playing time. Who should get minutes (and when) may be the coach’s most important set of decisions, especially when choosing between similar but subtly different players. So, what causes playing time in the NBA?
To find out, I correlated 21 stats to minutes per game. The strongest relationship was with my overall rating metric, Player Production Average. In other words, coaches tended to give players with better overall production more playing time.
Looking strictly at box score stats, coaches leaned strongest to two categories that are really variations on each other — field goals made and points scored (per 100 team possessions). If you want playing time in the NBA, be good at scoring. The next strongest categories were made threes, made twos and assists.
The defense portion of PPA correlates with minutes per game at the level of made threes, by the way. This suggests that coaches league-wide lean towards offense but also care about defense.
The areas that correlated weakest with minutes: offensive rebounding, fouls, blocks, turnovers, and defensive rebounding. A project for another time would be to compare this finding to previous seasons. These are all areas where bigs tend to produce larger numbers. Does this reflect the league’s trend toward playing smaller, more skilled players, or has this been consistent over time? Open issue for now.
Broadly speaking, the correlations tell us why players get minutes. The next question: what minutes allocations correlate best to winning? And, the strongest correlations between strength of schedule adjusted scoring margin and how coaches allocate minutes are: offensive rebounding, PPA, defensive rebounding, and the defense part of PPA.
A couple different interpretations spring quickly to mind. Perhaps the small-ball trend is leaving more productive players on the sideline. Or, perhaps coaches are a bit more likely to give playing time to high-quality big men. It could also not mean much at all — the statistical relationships are weak.
What seems to earn playing time with Brooks and the Wizards this season? Answer: field goal attempts, especially two-point attempts. So far this season, Washington is 25th in the relationship between PPA and minutes, and number one (by a lot) in the relationship between two-point attempts and playing time.
To earn minutes from Brooks this season, take lots of shots, especially if they’re twos. Bonus points if you can get to the free throw line. Extra bonus points if you get assists, grab rebounds, and block shots. Do your best to avoid fouling and turnovers, though.
One problem with reading too much into this approach is the issue of exactly how much choice a coach has. Brooks, for example, didn’t pick the players — Tommy Sheppard and the front office did. While Brooks participated in the player selection process, he didn’t have final decision power. He might prefer to allocate minutes differently but is limited by who’s on the roster.
Bradley Beal, Russell Westbrook, and Rui Hachimura are all guys with a penchant for pulling the trigger on two-point jumpers. They’re also each considered key players on the team. On offense, the coaching staff wants the team to shoot when open, including two-point jumpers.
The last item for today is to roll all this up into a minutes allocation score (with the understanding that this NOT a definitive measure of coaching prowess, and acknowledgement that due to injuries, roster composition, and other considerations, coaches may not have full control over playing time).
Brooks and the Wizards rank 10th in minutes allocation. In other words, Brooks may be doing a decent job deciding who plays and who doesn’t — at least in comparison to his coaching peers. The “SCORE” below is a calculation using the correlations between individual stats and playing time for each team, weighted by the relationship to winning.
Keep in mind that even this roll-up metric has only a loose (0.47 correlation) relationship with team success. This is in part likely because of differences in the quality of players. For example, the Detroit Pistons, a team with a losing record, rank third “best” in this metric. That could be an example of a coach tending to get his best players on the floor, but those players just not being very good.
Here’s the full list for this season so far:
NBA Minutes Allocation Scores