The “Stevens Theory” posits that coaching has an impact on NBA players. Specifically, they tend to perform better under Brad than they do before or after as a result of the “Stevens Effect.”
See table nearby.
Some notes about the table: Some players were quite young (Kelly Olynyk) or old (Amir Johnson) in their career when they came to play for Stevens. Or, when Stevens took over as coach (Avery Bradley.)
In general, we tried to account for two years “Before Stevens” and up to two years “After Stevens” wherever possible.
We left off Bradley’s rookie year, for instance, given he only played 162 minutes for the season. And even though he developed and improved significantly under Stevens’s tutelage, we included all four years following the departure of Doc Rivers.
And it’s only fair to note that in the case of some players, such as IT, the post-Stevens sample size is “solitary, brutish, and short.” As his hip recovers and team-mojo situation improves, so, in all probability, will his numbers.
(Don’t laugh too hard; it hurts. Since January 1, the Lakers have one of the best records in the league.)
As well, we used three different statistical measurements for performance. If your favorite one doesn’t appear, you’re welcome to @ us on twitter or the comments section below.
Some quick observations
You can see from the table itself what years we chose to compare, and draw your own conclusions. That being said, here are a few initial takeaways.
— The “Stevens Effect,” if any, finds its most cogent support from players like Isaiah Thomas, Jae Crowder, and Avery Bradley. There’s a decent amount of data on them before Stevens coached them, and at least a limited sample afterwards.
It also seems to work best with wings and point guards. This is not surprising. Stevens and Danny Ainge both seem to be most comfortable drafting, developing, and acquiring these. It would be interesting, of course, to look at a much larger database, something we may do in the future.
— Jeff Van Gundy remarked recently, “there are no bad coaches in this league; only bad rosters.” The Stevens Theory posits that, even if there are no “bad” coaches, there are “better” and “less better” coaches, let’s leave it at that.
But it’s important to note that most of the data about player movements, at least in the last few years, could just as well be attributed to Ainge. “It’s better to trade a player one year too early than one year too late,” and all that. In our opinion, any valid trends that do emerge involve a mixture of both Stevens and, of course, the “Ainge Effect.”
— A couple of redditors — Blackmanwdaplan and AmazonBrainForest — deserve acknowledgement. Their recent comments in a thread on Greg Monroe gave us the idea to look into this a little more systematically.
“I prefer RPM” and all that
If anything, there’s an interesting implicit comment on the statistics themselves. Given that many of the advanced stats purport to measure player’s overall contribution, it’s notable that some of them tell a very different story — for the same player.
Win shares, for example, are more sensitive to minutes played. So for a player like Morris, who’s seeing less time on the floor than in Detroit, there’s a natural drop-off. (His win shares per minute are actually up about 20 percent with the Celtics compared to his last two years with the Pistons.)
That’s why we went with several stats. Three data points are better than two.
Conclusion: To put it with scientific reserve, the Brad Stevens Theory has not yet been disproven. Further tests may be warranted. (Note: No current or former Celtics were hurt or mistreated during this experiment.)
The bottom line is, Brad Stevens appears to be a pretty good coach. (cc: Agents for 2018 and 2019 NBA free agent class.) How’s that for a hot take?
Editor’s note: The Stevens Theory is new, under development and under discussion, i.e., this is an ongoing series. Part one, a look at Shane Larkin’s stats and development, here. Part three, coming soon. Join in the discussion below, on the Celtics subreddit here, or on twitter at #StevensTheory.