|
For better or for worse, the Coach of the Year race basically comes down to which team most exceeded the media's preseason expectations, which is usually synonymous with "the team whose record improved the most over last year".
Take last year, for instance. The Bulls won 75.6 percent of their games under Tom Thibodeau, a 25.6 percentage-point gain over their .500 record under Vinny Del Negro in 2010. That represented the biggest improvement made by any team in the NBA, and Thibodeau won Coach of the Year honors as a result. It's nothing new--the coach with the biggest year-over-year improvement also won in 2010 (Scott Brooks), 2007 (Sam Mitchell), and 2005 (Mike D'Antoni), with the award going to the league's second-biggest improvers in 2008 (Byron Scott) and 2009 (Mike Brown).
So it's pretty much established now that Coach of the Year = "Coach of the team who surprised us the most." In 2004, John Hollinger (writing for SI at the time) attempted to quantify this by looking at not just the team's record in the previous year, but also two years prior (with a regression-to-the-mean component included for good measure). Here's what John had to say:
"To determine how many games the coach was expected to win, I pieced together a simple formula: The team's winning percentage the previous season (two parts), the team's winning percentage from two seasons previous (one part), and a .500 season (one part). I included the .500 season to adjust for the fact that a coach with 60 wins every year for a decade would otherwise show up as "+0." Similarly, a coach who went 10-72 every year would have the same mark even though he was awful. Because the draft and the salary cap conspire to level the playing field, it's a necessary addition. Finally, one other caveat--I only used the coach's record from seasons where he coached at least 50 percent of the team's games.
Using this formula, we can evaluate over time how a coach has fared against the expectation. For example, [Larry] Brown won 54 games with the Pistons last season, after they had won 50 the previous two seasons. Based on the formula, he was expected to win 49 games, so he gets a +5 for winning 54. Summing those up over all 24 seasons of Brown's career, we find that he's +114, or about five games better than expected each season."
Trouble is, both the "compare to last year" method and Hollinger's set expectations using the team's performance in Year Y-1 (and Year Y-2) as a proxy for talent in Year Y. That's not always true, as the leaderboards for both systems will tell us this year:
"Look at Last Year" Method
Coach Team W L WPct Y-1 +/-
-------------------------------------------------------------
Vinny Del Negro LAC 38 23 0.623 0.390 0.233
Frank Vogel IND 40 22 0.645 0.451 0.194
Rick Adelman MIN 25 38 0.397 0.207 0.190
Byron Scott CLE 20 40 0.333 0.232 0.102
Dwane Casey TOR 22 40 0.355 0.268 0.087
Avery Johnson NJN 22 40 0.355 0.293 0.062
Scott Skiles MIL 29 31 0.483 0.427 0.057
Larry Drew ATL 36 25 0.590 0.537 0.054
Scott Brooks OKC 44 17 0.721 0.671 0.051
Keith Smart SAC 18 36 0.333 0.293 0.041
Coach Team W L WPct Y-1 +/-
-------------------------------------------------------------
Tyrone Corbin UTA 32 30 0.516 0.476 0.041
Alvin Gentry PHO 32 29 0.525 0.488 0.037
Lionel Hollins MEM 36 25 0.590 0.561 0.029
Randy Wittman WAS 13 31 0.295 0.280 0.015
Lawrence Frank DET 23 38 0.377 0.366 0.011
Erik Spoelstra MIA 43 17 0.717 0.707 0.009
Doug Collins PHI 31 30 0.508 0.500 0.008
Kevin McHale HOU 32 29 0.525 0.524 0.000
Tom Thibodeau CHI 46 15 0.754 0.756 -0.002
Gregg Popovich SAS 44 16 0.733 0.744 -0.011
Coach Team W L WPct Y-1 +/-
-------------------------------------------------------------
Stan Van Gundy ORL 36 25 0.590 0.634 -0.044
George Karl DEN 34 27 0.557 0.610 -0.052
Mike Brown LAL 39 23 0.629 0.695 -0.066
Mark Jackson GSW 22 38 0.367 0.439 -0.072
Mike D'Antoni NYK 18 24 0.429 0.512 -0.084
Doc Rivers BOS 36 26 0.581 0.683 -0.102
Nate McMillan POR 20 23 0.465 0.585 -0.120
Rick Carlisle DAL 34 28 0.548 0.695 -0.147
Monty Williams NOH 19 42 0.311 0.561 -0.250
Paul Silas CHA 7 53 0.117 0.415 -0.298
Hollinger Method
Coach Team W L WPct Y-1 Y-2 Exp % +/-
-------------------------------------------------------------------------------
Vinny Del Negro LAC 38 23 0.623 0.390 0.354 0.409 0.214
Frank Vogel IND 40 22 0.645 0.451 0.390 0.448 0.197
Tom Thibodeau CHI 46 15 0.754 0.756 0.500 0.628 0.126
Rick Adelman MIN 25 38 0.397 0.207 0.183 0.274 0.122
Scott Brooks OKC 44 17 0.721 0.671 0.610 0.613 0.109
Erik Spoelstra MIA 43 17 0.717 0.707 0.573 0.622 0.095
Gregg Popovich SAS 44 16 0.733 0.744 0.610 0.649 0.084
Lionel Hollins MEM 36 25 0.590 0.561 0.488 0.527 0.063
Doug Collins PHI 31 30 0.508 0.500 0.329 0.457 0.051
Avery Johnson NJN 22 40 0.355 0.293 0.146 0.308 0.047
Coach Team W L WPct Y-1 Y-2 Exp % +/-
-------------------------------------------------------------------------------
Larry Drew ATL 36 25 0.590 0.537 0.646 0.555 0.035
Kevin McHale HOU 32 29 0.525 0.524 0.512 0.515 0.009
Scott Skiles MIL 29 31 0.483 0.427 0.561 0.479 0.005
Tyrone Corbin UTA 32 30 0.516 0.476 0.646 0.524 -0.008
Alvin Gentry PHO 32 29 0.525 0.488 0.659 0.534 -0.009
Lawrence Frank DET 23 38 0.377 0.366 0.329 0.390 -0.013
Keith Smart SAC 18 36 0.333 0.293 0.305 0.348 -0.014
Mike Brown LAL 39 23 0.629 0.695 0.695 0.646 -0.017
Dwane Casey TOR 22 40 0.355 0.268 0.488 0.381 -0.026
Stan Van Gundy ORL 36 25 0.590 0.634 0.720 0.622 -0.032
Coach Team W L WPct Y-1 Y-2 Exp % +/-
-------------------------------------------------------------------------------
George Karl DEN 34 27 0.557 0.610 0.646 0.591 -0.034
Doc Rivers BOS 36 26 0.581 0.683 0.610 0.619 -0.038
Mike D'Antoni NYK 18 24 0.429 0.512 0.354 0.470 -0.041
Randy Wittman WAS 13 31 0.295 0.280 0.317 0.345 -0.049
Mark Jackson GSW 22 38 0.367 0.439 0.317 0.424 -0.057
Rick Carlisle DAL 34 28 0.548 0.695 0.671 0.640 -0.092
Byron Scott CLE 20 40 0.333 0.232 0.744 0.427 -0.093
Nate McMillan POR 20 23 0.465 0.585 0.610 0.570 -0.105
Monty Williams NOH 19 42 0.311 0.561 0.451 0.518 -0.207
Paul Silas CHA 7 53 0.117 0.415 0.537 0.466 -0.350
That's right, Vinny Del Negro, who darn near got himself fired in March, would be your 2012 Coach of the Year by these tried-and-true methods of setting up coaching expectations.
Obviously, there has to be a better way of accounting for the raw talent on a team beyond simply using the franchise's previous W-L. And that's where Daniel Myers' Advanced Statistical Plus/Minus comes in. In essence, ASPM is a boxscore-based, more stable version of Regularized Plus/Minus that is probably the most predictive all-in-one advanced box-score stat out there right now. Denominated in efficiency differential, ASPM is ideal for projecting a team's performance via a minute-weighted average of the individual ratings of its players.
To set a team's expectations for 2012, I looked at each 2012 player's ASPM score from 2011 (using Daniel's latest spreadsheet). If a player played 500+ minutes in 2011, his projected ASPM for 2012 was his 2011 ASPM. If he played less than 500 minutes in 2011 (and isn't a 2012 rookie), he was given a constant -3.20 ASPM projection (the minute-weighted average of all low-minute non-rookies in 2011); if he is a 2012 rookie, he was given a constant ASPM of -1.54 (the minute-weighted average of all 2011 rookies). I then used these "projections" and each team's actual 2012 playing-time distribution to create an expected efficiency differential for the team in 2012.
Comparing the weighted average of their actual 2012 ASPMs to that expected differential, then, provides a measure of how much a team (and a coach) exceeded expectations, given the amount of talent on his roster. For instance, here are the expected and actual ASPMs for New Orleans in 2012:
Player MP Proj Actual
-----------------------------------------
Marco Belinelli 1839 -1.69 -1.75
Greivis Vasquez 1545 -4.09 -0.43
Jarrett Jack 1530 -1.62 1.05
Chris Kaman 1372 -1.32 -1.64
Trevor Ariza 1350 0.05 0.78
Al-Farouq Aminu 1305 -3.04 -1.64
Gustavo Ayon 1004 -1.54 1.88
Carl Landry 889 -1.45 -1.14
Jason Smith 873 -2.54 -0.44
Emeka Okafor 781 1.14 0.03
Xavier Henry 694 -4.44 -2.93
Lance Thomas 495 -1.54 -2.81
Eric Gordon 234 1.36 2.13
DaJuan Summers 209 -3.20 -2.95
Solomon Jones 196 -3.09 -2.83
Squeaky Johnson 119 -1.54 -2.78
Chris Johnson 82 -3.20 -1.88
Trey Johnson 61 -3.20 -3.11
Jerome Dyson 56 -1.54 -4.69
Donald Sloan 41 -1.54 -5.61
Jeff Foote 39 -1.54 -6.54
-----------------------------------------
Team -9.32 -3.61
+5.71
In fact, that +5.71 mark was the NBA's best. Given the talent of its players, New Orleans "should" have had a schedule-adjusted efficiency differential of -9.3, which, while not Charlotte-level bad, would have been well clear of Cleveland and Washington for the league's second-worst mark. Instead, the Hornets rank 23rd--still below-average, so not great by any means, but much better than you would have thought before the season if you looked at the players Monty Williams had to work with.
So by this metric, here is a truer ranking of the league's most excellent coaches:
Coach Tm Proj Actual Diff
----------------------------------------------------
Monty Williams NOH -9.32 -3.61 5.71
Tyrone Corbin UTA -4.79 0.24 5.04
Larry Drew ATL -2.71 2.27 4.98
Gregg Popovich SAS 1.85 6.62 4.77
Frank Vogel IND -0.72 3.06 3.77
Tom Thibodeau CHI 4.78 7.85 3.07
Rick Adelman MIN -4.43 -1.70 2.74
Dwane Casey TOR -6.74 -4.06 2.67
Scott Brooks OKC 4.20 6.87 2.67
Doug Collins PHI 1.79 4.20 2.41
Kevin McHale HOU -1.61 0.74 2.35
George Karl DEN 0.25 2.30 2.05
Alvin Gentry PHO -1.44 0.58 2.02
D'Antoni/Woodson NYK 0.97 2.70 1.73
Rick Carlisle DAL 0.38 1.85 1.47
Mark Jackson GSW -3.69 -3.06 0.64
Saunders/Wittman WAS -7.51 -7.00 0.51
Erik Spoelstra MIA 6.41 6.86 0.45
Lionel Hollins MEM 1.89 2.29 0.40
Byron Scott CLE -7.73 -7.45 0.28
Scott Skiles MIL 0.27 0.24 -0.04
Lawrence Frank DET -5.31 -5.53 -0.22
Vinny Del Negro LAC 3.84 3.16 -0.68
Westphal/Smart SAC -4.81 -5.79 -0.99
McMillan/Canales POR 1.53 0.54 -1.00
Doc Rivers BOS 3.56 2.11 -1.45
Avery Johnson NJN -3.82 -6.03 -2.21
Stan Van Gundy ORL 4.25 1.53 -2.72
Mike Brown LAL 5.99 2.31 -3.68
Paul Silas CHA -7.30 -14.28 -6.97
I think that's a fairer metric because it more accurately accounts for what a coach had to work with. And I suppose if you really want to make it more accurate, the "projections" could factor in a greater number of past years, and an aging adjustment could be added.
At any rate, a method like this represents a better way of assessing which coach truly squeezed the most out of his players. Perhaps Monty Williams doesn't deserve Coach of the Year on the basis of his team still being bad (just less bad than it could have been), and you might hold it against Corbin if the Jazz miss the playoffs, but it's tough to argue that Drew, Popovich, Vogel, and Thibodeau don't have their respective teams playing much better than we'd expect from how their players performed a year ago.
This free article is an example of the kind of content available to Basketball Prospectus Premium subscribers. See our Premium page for more details and to subscribe.
Neil Paine is an author of Basketball Prospectus.
You can contact Neil by clicking here or click here to see Neil's other articles.
| 
Three Centers (04/17)
0 comments have been left for this article. 
