This season, Derrick Rose has not been his old MVP self. This is not a revelation, I’m not telling anyone noteworthy information by making that observation. Rose has shown flashes of brilliance here and there, but mostly he’s been okay to pretty bad depending on the night. His overall numbers and efficiency have all trended down over the course of the year. I’m still not quite ready to panic regarding Rose’s struggles, though. Some of this is based on positive things Rose has been able to do and part of it might be wishful extrapolation from a select few (some would say cherry-picked) numbers. Let’s get into it.
Pau Gasol was selected as an All Star starter by the fans. This is, honestly, a joke. Pau Gasol is the Bulls fourth best big man. Yes, fourth best. Behind Joakim Noah, Taj Gibson, and even rookie Nikola Mirotic. The idea that Gasol was somehow one of the best big men in the entire Eastern Conference is ridiculous. At first glance, it’s easy to see why Pau has received this recognition. Pau is averaging 18 points, 12 rebounds, nearly 3 assists, and over 2 blocks in his roughly 35 minutes per game. Those certainly seem like All Star numbers. However, it’s pretty likely that no one is putting up emptier stats than Pau Gasol is this season.
It’s been a while since I’ve written anything in this space, but I’ve not forgotten it, and I’ve not stopped tinkering with some of the ideas I’ve tried to tackle here before. The idea I’d like to return to today is my enjoyment of simple, easy to calculate, transparent boxscore metrics.
In the past, I built off of an easy to calculate and understand linear weights metric (Alternate Win Score) to create Usage Adjusted Rating, which essentially tried to adjust AWS to credit heavier usage players for the greater degree of difficulty they generally encounter in getting points and remaining efficient. The results were pretty good and passed the laugh test. But calculating UAR and the subsequent variant blend with plus-minus (UARPM) that I developed was best done on season long numbers and well, there are much better one number metrics out there for analysis of season long data. Daniel Myer’s Box Plus-Minus (BPM) and ESPN’s Real Plus-Minus (RPM) being the best examples. So from here on, I’ll be retiring UARPM from the website.
But for broad strokes analysis of single games, the current best linear weights metric is probably Alternate Win Score. Some people like to use John Hollinger’s Game Score, since it’s readily available on Basketball Reference for every game. I wanted to improve upon AWS and Game Score and build a transparent, easy to calculate and understand metric to quickly analyze game to game performance.
To build my game score metric, I looked to Jerry Engelmann’s 14 year RAPM data set, since it is, to my mind, the best estimate of long run +/- impact that’s in the public domain. After that, I ran a regression of the most basic boxscore stats (per 100 possessions) to get the coefficients or weights for my simple linear weights metric.
I tried to include personal fouls, but they were not statistically significant predictors of RAPM (+/-) at all. All of the other boxscore stats I picked were highly statistically significant with strong p-values. Then, I translated the coefficients so that they were weighted relative to points score (i.e., so that the coefficient or weight for points was equal to 1). The resulting weights for my simple game score metric are as follows:
PTs + .2*TRBs + 1.7* STLs + .535*BLKs + .5*ASTs — .9*FGA — .35*FTA — 1.4*TOV
If you want to translate this linear weights metric directly to a simple statistical plus-minus, you can just subtract the average performance league-wide from the player’s total. Per pace adjusted 36 minutes the average performance in the league currently is roughly 4.9. Here’s the top 25 in the league as of the games played January 16 per game, pace adjusted, with the per-minute average subtracted out, so as to make it roughly +/- impact per game:
PlayersPM/GAnthony Davis7.1Stephen Curry5.9Chris Paul5.7James Harden5.0Kevin Durant4.3Jeff Teague3.9Jimmy Butler3.8LeBron James3.7Kyle Lowry3.5Damian Lillard3.3Russell Westbrook3.1DeAndre Jordan2.9Tyson Chandler2.8Brandan Wright2.6Kawhi Leonard2.6Kevin Martin2.6John Wall2.6Mike Conley2.5Klay Thompson2.4Paul Millsap2.3Marc Gasol2.3Hassan Whiteside2.3Derrick Favors2.2Kyrie Irving2.1Kyle Korver2.1
Those results definitely pass the laugh test. Anyway, I like this as another tool in the tool kit. I even won over noted one number metric skeptic Seth Partnow to use the metric for some broad strokes performance analysis.
Good enough for me!
If any of you have a good idea for a name for this new Win / Game Score linear weights metric, let me know in the comments or on Twitter: @NBACouchside.
Let me get this out of the way at the start: I love Mike Dunleavy Jr. He’s really, really good. You should love Mike Dunleavy Jr., too. Here’s why:
Through 4 games, Tom Thibodeau’s defense is outside of the league’s top 10 in points allowed per 100 possessions and that is, obviously, very surprising. Granted, Chicago is 11th in defensive efficiency, so they are barely outside of the top 10, but given that under Thibodeau the Bulls have finished 1st, 1st, 5th, and 2nd in overall defense in the last 4 seasons, it is a bit surprising to see them anywhere but the very tip top of the league’s defensive rankings, even at this early juncture.
The subject of this article might not seem the most timely, given that Pau Gasol just beat the stuffing out of the New York Knicks in the Chicago Bulls’s season opener, recording an efficient 21 point, 11 rebound night in just 29 minutes of play. In fact, this might all end up being a bit of nagging worry that doesn’t amount to much of anything at all. But although the Knicks’ interior defense is incredibly bad, there were some not so great signs for Pau’s prognosis in that game, ones that were also evident during the pre-season as well.
Like Andrew Johnson, I did some season projectin’ for the season starting tonight. The basic method was to take ESPN’s Real Plus-Minus (RPM) numbers and run them through a simple aging curve and then to project out the season minutes.
[Ed. note: This piece originally ran on July 23, 2014 on a site where all traces of my former presence, along with that of all of the other members of the team, has been wiped away. I didn’t want it to disappear from the internet, so I’m reposting it here, where I have control over what happens to it.]
I am dumber than I like to think that I am. From time to time, it’s important to remind myself of this. This article is me doing just that.
It is often easy when writing about sports to fancy yourself just as qualified as anyone else, including say general managers and talent evaluators, to say whether a player might be bad, average, good, or great. This is, for the most part, harmless, and in some cases, it is possible to outperform the average or bad general manager, if you’re pretty good at scouting talent (like all of the armchair GMs among us would like to believe we are). But this attitude can morph into the worst kind of narcissistic hubris, and well, make you (and when I say “you,” understand that I mean “me”) come off as a bit of an ass. Now, front office types do not have a corner on the market for knowledge about the game, and this should not read like an argument or claim that they do. What I’m wrestling with is that none of us has a perfect understanding of the game, and we’re all learning new things every day we’re lucky enough to watch the world’s other beautiful game. It’s easy to be blind to that. It’s so easy, especially in the super overreactionizer that is Twitter, to have a strong reaction to something and spew it out without really stepping back to question your own assumptions.
It’s quite hard to challenge those assumptions regularly; it’s so much easier to allow them to calcify and constrict my thinking. It requires no effort at all to fall back on my default setting: I know this and that and this too about basketball. But, then, every once in a while, I get to take a moment and breathe and really think, and what I always come back to is this quote from Socrates:
I know one thing: that I know nothing.
It is probably the most important sentence I’ve ever heard about the nature of knowledge. I should always be striving to learn and understand better. Whenever I decide I know something, I’m lost, because I’ve stopped learning.
This is all a bit abstract, so let me be more concrete. When the Chicago Bulls, my favorite basketball team, traded a bunch of assets to acquire Creighton star Doug McDermott, I basically had a Twitter meltdown. McDermott is probably one of the best shooters in the world on a team that was terrible at scoring last year, but all of my favorite statistic-based models cast lots of doubts about whether Doug could play at the NBA level. He didn’t pass very much, he basically never got blocks or steals, and his rebounding was merely decent. McDermott’s low block and steal rates and just okay rebounding made me worry about his athleticism, as those three stats have traditionally been pretty reliable at predicting which players will have the athleticism to hang in the league and those who won’t. The concerns about McDermott’s athleticism matched my own eye test concerns about him. So I decided I knew who Doug McDermott was as an NBA player before ever seeing him play in the NBA. I ignored people, like my friend Ricky O’Donnell of BlogaBull, who pointed out Creighton’s ultra-conservative defensive scheme as a reason for his low defensive counting stats. I scoffed off people who told me he was a good body position defender. I disregarded the common-sense idea that when you’re scoring as much as McDermott did and moving all over the court non-stop with literally five defensive players all aimed at stopping you, maybe defense takes a bit of a backseat. I stopped thinking and started ranting. I was lost.
The statistical models I love have had a pretty good success rate, especially when compared with the average general manager in the NBA, but they’re not infallible. There are plenty of misses, just as there are with any attempt to predict the future of very young men making the transition to a totally new atmosphere and level of competition. There are simply too many things we can’t know at the time of the draft which effect how well a player will do at the next level. How they played in college or internationally and their resulting counting stats is a big piece of that puzzle, but it is only part of it. I ignored that, too.
I watched Doug McDermott in summer league, and many of my concerns still linger. He probably won’t score as prolifically as he did in college and his lateral quickness is not great. But McDermott is so, so smart. He makes extremely quick decisions with the ball, which is a still undervalued skill and its value is multiplied exponentially by the threat he represents as a shooter and floor spacer, especially given his lightning quick shooting release. He’s going to bend defenses, just by virtue of these two skills. He’s also a much better passer and decision-maker than his low college assist totals would suggest. Additionally, McDermott is, as I was told, a solid body position defender, who will mostly funnel players towards his help defense- perhaps not coincidentally, the Bulls have two of the league’s very best help big men in Joakim Noah and Taj Gibson. Yes, McDermott will give up blow-bys to more athletic players and yes, that will be frustrating when it happens, but it won’t hurt as much as it might on another team because of those two big mobile guys behind him. Context matters very much in basketball, and well, maybe on draft night and after I didn’t think enough about the context in which McDermott will be operating. The defensive warts can be more easily hidden in Chicago than nearly anywhere else and his skill-set is a much needed one on any team, but especially for these Bulls.
Film Crit Hulk is one of my absolute favorite writers, and he has a tremendous piece which centers around a bit of advice given to him by the famed director, Quentin Tarantino. During a conversation in which a younger, perhaps less thoughtful Hulk ranted against a movie he “hated,” Tarantino told him, “Never, under any circumstances, hate a movie. It won’t help you and it’s a waste of time.” Tarantino went on to more fully explain that there is value and things to learn and enjoyment to be found even in the bombs or, for our basketball-watching purposes, busts. Tarantino finished his advice by saying of movies, “They’re gifts. Every f*cking one of ‘em.”
I am much less sure than I was about what sort of player Doug McDermott might be than I was on draft night. Part of that is a function of his summer league play, but a bigger part of it is me allowing myself to embrace that I don’t know nearly as much as I sometimes think and act like I do. What I do know is that regardless of whether he turns out to be a “bomb” or another “hit” for the Bulls front office, I’ll learn from watching him play. I’ll learn from seeing his struggles or successes and the how and why behind them. I’ll be entertained, as I always am, by the process. Doug McDermott is a gift, just like every player which I have the privilege to watch and root on.
Stan Van Gundy was, to my mind, the best coach available on the market this offseason, and for the last few offseasons frankly. That stance, of course, comes with the caveat that there are many assistant coaches and coaches at different levels of basketball with whom I am unfamiliar and therefore, I exclude from any consideration in the “best coach available” discussion. There may be some diamonds in the rough who need a shot to show how good they can be. But I don’t know enough about the coaching development chain to say much about those guys. Still, SVG has proven himself as a great coach in the NBA.
Van Gundy ought never to have lost his job in the first place, but it’s a superstar league and the Magic wanted to appease Dwight Howard, before realizing how futile a game that really was. SVG has a winning percentage of 64.1% in the regular season over 579 games and he is 9 games over .500 in the post-season (48–39). Winning percentage alone, however, is not the only mark of a great coach. If you need any proof of that, check out Avery Johnson’s career after his Dallas turn, Vinny Del Negro’s run with the Clippers last season, or Scotty Brooks’ campaigns the last few seasons. Talent can overcome imperfect or even downright incompetent coaching, at times. Van Gundy isn’t that sort of coach. He’s exceedingly competent, curious, and supremely adaptable. So the following should hardly be surprising:
Stan Van took two separate and totally different style squads through deep playoff runs, ending in the Conference Finals and NBA Finals, respectively. In Miami, he paired Dwyane Wade’s pell-mell drives to the hoop with a still dominant Shaq’s low-post scoring to reach a top 5 offense and a 7 game conference finals loss to Detroit. It was a distinctly traditional offensive setup, and Van Gundy made it work incredibly well.
Then, from 2008–2011, he helped Dwight Howard reach his peak, which he has still not surpassed or even really approached, with a 4-out spread pick and roll attack which both incorporated the biggest lessons of the modern basketball analytics movement (the importance of spacing, along with the value of the three point shots and dunks) and best catered to the talents of his team. It’s also worth noting that no coach since SVG has been able to get Dwight to buy in to the pick and roll game as thoroughly as he did in Orlando. In 2009, the Magic overachieved in the playoffs, knocking off LeBron’s Cavaliers juggernaut — +8.68 SRS, well ahead of the second best team, the eventual champion Los Angeles Lakers — in the Eastern Conference Finals, only to get beat by that Lakers team in 5 games. It would be easy to be disappointed by the Magic’s performance in that Finals, but really, they shouldn’t have even been in that series at all. Orlando had already overachieved by reaching the Finals. Of course, that’s not the attitude a coach or his team should have in that series, but for fans and media analyzing Van Gundy’s resume, it’s important to note that important context.
So, in short: Detroit has nabbed a great coach. Van Gundy, after being burned by both his former Florida employers and unjustly let go, also managed to negotiate control over Basketball Operations for the Pistons. I don’t know how SVG will do on the management side of things, as there’s no history to go on. Given his understanding of how teams fit together and how to maximize the talents he’s provided, though, I suspect he’ll also do a heck of a job there, as well. He would be hard pressed to do worse than Joe Dumars the last few years.
I see most things in the NBA through the lens of how they will effect my beloved Chicago Bulls. Stan Van Gundy entering the Central Division scares the hell out of me. I’d imagine most fans of other Central Division teams feel likewise, which is as good an indication as any that Detroit has done very well for themselves here. Now it’s time to fire up the trade machine, because there’s almost no way SVG is keeping this trainwreck together next season.
Last post, I mentioned that Kevin Durant was the UARPM100 MVP, and I gave a top 10 list of players in Wins Above Replacement as well. After looking through the numbers, something that occurred to me was that the number of total wins under those numbers didn’t sum up to team level wins. That was primarily an effect of including raw per minute plus-minus numbers as part of the UARPM formulation. Basically, good teams had too many wins, and poor teams had too few wins. So I decided to correct that. I adjusted the UARPM100 numbers using a per minute adjustment for each player on the team so that total team plus-minus was equal to team SRS (basically point differential adjusted for strength of schedule) via Basketball-Reference. The final numbers are posted on the UARPM100 page.
The top 10 is basically the same, with Carmelo Anthony jumping into the 8th spot, and DeAndre Jordan sliding to 10th. The total number of wins are reduced across the board, and they are no longer set to above replacement, because I decided it’d be more interesting to just have total wins contributed. You can easily turn wins into Wins per 48 minutes by dividing by minutes played and multiplying by 48. Durant and LeBron were nearly exactly the same in per possession impact by UARPM100, with Durant’s heavier minutes load giving him the edge in wins. Chris Paul also remains the best per-possession player in the league, even after the team adjustment.
Here’s the updated top 20:
RankPlayersMINUARPM100Wins1Kevin Durant31226.720.02LeBron James29026.718.63Kevin Love27966.417.64Stephen Curry28466.117.35Blake Griffin28635.215.86Chris Paul21717.214.77Joakim Noah28204.614.38Carmelo Anthony29824.114.39James Harden27774.514.110DeAndre Jordan28704.213.911Paul George28984.113.912DeMarcus Cousins22985.913.713John Wall29803.513.114LaMarcus Aldridge24994.713.015Kyle Lowry28623.512.616Dwight Howard23964.612.217Goran Dragic26683.611.918Anthony Davis23584.411.719Al Jefferson25533.811.720Serge Ibaka26663.411.6
All in all, this seems like a pretty credible list. For what it’s worth, Rookie of the Year award winner, Michael Carter-Williams produced 6.1 wins under UARPM100, well ahead of runner-up Victor Oladipo who clocked in at 4.7 wins. The voters appear to be doing a pretty good job.