With a record 11 championships to his credit, it’s easy to argue that Phil Jackson is the greatest coach in NBA history.
Then again it would be hard not to rack up the rings coaching Michael Jordan, Scottie Pippen, Kobe Bryant, Shaquille O’Neal and Pau Gasol.
That’s not to take away from Jackson’s greatness as a coach — few coaches in league history could so adeptly manage all those personalities — but it’s a fact that the NBA is a player’s league and talent often trumps all else.
So that begs the question, how much do coaches matter in the NBA?
Clearly Jackson is a brilliant people manager and with the help of Tex Winter a superb tactician, but how many rings would an average coach have won with Jordan, Pippen, Shaq, Kobe and Gasol?
Jeremias Engelmann’s site “Stats for the NBA” attempts to answer the gist of that question — what kind of a difference do coaches make? — by analyzing how players perform in contrast to expectations with different coaches.
According to Engelmann: “The system gives estimates for players. If the players perform better than their estimates under a certain coach the coach will look good, even if all his players are bad players. A coach can also get a bad rating if his players perform worse than what the system expected them to. Carlisle in Dallas isn’t seen as a good offensive coach by the algorithm even though Dallas is a good offensive team. The system expected them to be better, because they were a better offensive team under other coaches.”
So basically the system compares how players performed under previous and future coaches with how they performed with the coach at hand to see how the coach improved or hurt the team’s respective offense and defense. So just because the Suns’ offense has been elite in recent years and its defense below average doesn’t necessarily mean their coaches will rank thusly since it’s more a comparison to expectations.
Tom Liston from Raptors Republic wrote of the system: “The easiest way to interpret this analysis is this: it attempts to treat the coach as if he is the ’6th man’ on the court. It’s measuring the coach’s influence on those who play, but as Mr. Englemann highlights ‘it can’t tell you whether a coach is good at identifying who is good and who is bad, and if he plays the good players more because of that.’”
On to the Suns, Alvin Gentry ranks 39th and grades out slightly negative overall with his teams playing -0.9 points per 200 possessions worse than expected. His offenses are a bit better than expected (positive 0.5) and defenses worse (-1.3). Keep in mind this data covers the period from 2002 to the present so the tail end of Gentry’s Clippers career is included.
Mike D’Antoni, 25th on this list of 97 coaches, barely impacts his teams at all according to this algorithm, with his squads performing -0.1 points per 200 possessions worse than expected. As one would expect, his offenses were above average (0.8) and defenses below average (-0.9).
Terry Porter not surprisingly ranked 1.0 points per 100 worse on offense but 0.3 better on defense to be 0.7 worse overall. It’s fair to say the Suns’ offense was a bit worse under his guidance than under that of D’Antoni or Gentry.
Going back a bit further, Frank Johnson’s teams were 2.2 worse than expected on offense but 2.0 better on defense.
Of course there are limitations to this analysis. It does not take aging into effect so, as Engelmann noted in an email, Scott Brooks will be credited with the natural improvement of Durant, Westbrook and the rest of the young Thunder players while John Kuester was negatively impacted by the demise of the Pistons.
Similarly a coach who got Shaq the last few years may be unfairly punished while Jackson will get more credit than he deserves just for coaching The Big Diesel during his prime.
However, this algorithm’s main conclusion is that the majority of coaches don’t make a huge difference. Only Tom Thibodeau improves his team by better than two points more than expected (and that should go down once he coaches more than one season), and just 10 coaches make their teams even a point better than expected. More coaches performed worse than expected (only 21 made a positive difference) but only five made their teams more than three points worse than expected.
This analysis does prove in part that coaches make their biggest impact on the defensive end as four of the top five coaches and five of the top eight led defenses more than three points better than expected. Former Suns coach Scott Skiles ranked as the best defensive coach (3.7 better per 100), while Thibodeau, Mike Fratello and Jeff and Stan Van Gundy round out the top five best defensive coaches (which passes the smell test).
It’s also worth nothing that new Suns defensive coach Elston Turner has spent the past decade coaching with Rick Adelman, whose defenses have performed 1.9 points better than expected despite his reputation as an offensive coach.
Tony DiLeo (1.4), Sam Mitchell (1.3) and Kevin McHale (1.2) rate as the best offensive coaches with D’Antoni tied for fourth and Gentry seventh. Former Suns coach Paul Westphal ranks eighth but only nine coaches rank better than 0.1, showing once again coaches can make a bigger impact defensively. UA fans might be interested to know former interim head coach Kevin O’Neill ranks last by a wide margin with an offense losing 4.1 points per 100 possessions than expected.
The great Phil Jackson himself ranks 11th overall (1.2 worse on offense and 2.1 better on defense for a 0.9 net positive), Jerry Sloan was 14th (0.3 worse offense, 0.8 better defense, 0,4 better overall) and Gregg Popovich 32nd (2.8 worse offense, 2.5 better defense, 0.4 worse overall).
Prof. David Berri of Wages of Wins fame has run a similar analysis using Win Score as his evaluation determinant and found that coaches don’t make a very big statistical difference.
Basketball Prospectus’ Kevin Pelton poked some holes in that analysis because there are so many different areas that a coach touches that it would be unfair to evaluate his effectiveness purely by player improvement.
My takeaway is that it is a bit unfair to judge coaches based on how effective their players play under them in comparison to how they play under other coaches because so many variables such as age and teammates cannot naturally be factored in.
As for the Suns it’s clear they have employed superior offensive coaches the past seven years aside from the Porter blip, although so long as we are asking questions it’s fair to ponder how much credit should be given to Steve Nash rather than the head coach for this offensive uptick.
It does appear as if the most effective coaches by this metric are known as defensive gurus, which lends credence to the theory that a coach can be the best possible “sixth man” by implementing a superior defensive game plan, which has not exactly happened in Phoenix this decade.
It would only be fair to conclude by noting there are so many other things that go into a coach’s job that do not factor directly into an analysis like this, such as crunch time execution and the ability to make a critical halftime adjustment and create chemistry.
Like many advanced stats this coaching stat isn’t perfect, especially to those who evaluate coaches based on rings, but it still provides an interesting window into how NBA coaches impact their teams.
Wages of Wins Journal on coaching
After this post was published, Andres Alvarez directed me toward Arturo Galletti’s post on the Wages of Wins Journal analyzing the impact of coaches.
Galletti graphed minutes per game against Wins Produced per 48 minutes for every player season since 1978 and found that “a player’s actual playing ability only accounts for less than a third of the variation we see in playing. What that means is that every time you scream at the television that the wrong guys are getting the playing time, there’s a good chance you are correct.”
As Alvarez cleared up for me on Twitter, this means a player’s ability (as seen through WP48) only explains one-third of minute allocation with the other two-thirds being “who knows.”
The “Stats for the NBA” coaching stat does not take into account who the coach plays, but this is clearly a major area in which coaches should be judged as it’s perhaps the biggest way they impact a game.
Galletti concludes by analyzing how playing times correlates with WP48, and Doug Collins won by a landslide last season with a 84.7 percent correlation. Gentry ranked 17th with a 38.9 percent correlation in part because of how much he played Vince Carter, Robin Lopez and Channing Frye, who puts up weak WP48 numbers.