At the Blogs With Balls conference that I attended last weekend, advanced analytics frequently came up, and was the topic of an entire panel. As a proponent for and user of “beyond the boxscore” metrics to evaluate basketball players, I needed no convincing, but a panelist on a hockey panel brought up something I thought was very interesting. He asserted that, while there were certainly sportswriters with their hands over their eyes refusing to understand new ways of assessing players, advanced statistics proponents were equally guilty of arrogantly writing as if they knew everything and everybody else was stupid.
Yesterday on HoopSpeak Live, in relation to a conversation about what the Thunder should do with James Harden, a debate broke out regarding Russell Westbrook’s true value. I won’t get into the full back-and-forth, but suffice to say that opinions on Russell’s ability ranged from “top 7” to “average”. I found myself guilty of what the hockey panelist had asserted, as I jumped into the debate armed only with a couple numbers and the assurance that I was right.
Jared Dubin, of Hardwood Paroxysm, objected to the structure of the debate, asking why we boil players abilities down to a single, all-encompassing number, instead of looking at those in context alongside other important metrics like Points per Possession, plus/minus, per-36 minute averages and so on. In an ideal world, we would have a model that perfectly encapsulates a players exact value, but absent that, Dubin has an important point. No one number should be taken as gospel.
There has been a lot of ink spilt arguing which all-encompassing metric is best. I’m not going to wade too deeply into that, especially because I will soon find myself drowning in the deep end of something I don’t understand, except to say that I am not an uninterested party. The three most widely used all-encompassing metrics, to my knowledge, are Wins Produced (Wages of Wins), Win Shares (Basketball Reference) and PER (John Hollinger, ESPN). Everything I have read leads me to believe that PER is a relatively worthless measure of a players’ ability, and Wins Produced is the best measure (that we currently have). I’ve also had a few posts up on the Wages of Wins blog. I’m not in any way a Wages of Wins acolyte, but I do believe they are doing good statistical work.
With that out of the way, let’s look at Russell Westbrook. I will do my best to evaluate his 2011-12 season holistically and attempt to determine where, right now, he ranks compared to other NBA players.
There is no doubt that Westbrook is a very talented player. As a point guard he led the league’s second most efficient offense, showcasing his ability to attack the basket, shoot, pass and rebound at elite levels. On defense, he has the quickness to stay with the smaller point guards and the strength to stand up to bigger ones. He’s strong enough that he doesn’t suffer too much if he is switched onto a shooting guard.
During the course of the offseason, ESPN ran its (second) annual NBA Rank project, gathering over 100 analysts to rank every single player in the NBA. These analysts ranked Westbrook the ninth best player in the league, which I think conforms pretty well with the popular perception of him. To the vast majority of NBA fans, there is no doubt that Westbrook is an elite player.
Wins Produced: 77 Wins Produced/48: 186 Win Shares: 13 Win Shares/48: 50 Plus/Minus: 96 PER: 12
In aggregate, the all-encompassing metrics aren’t as rosy on Westbrook’s ability than the sniff test would lead us to believe. One clear take away is that some of Westbrook’s high raw stats are due to minutes played: last season he played the seventh most minutes in the league. Both Wins Produced and Win Shares, which differ vastly on his ability, agree that he is worse on a per minute basis. Now, there is something to be said for Westbrook’s durability and high-level of play despite shouldering a heavy minute load, but it shouldn’t blind us to the fact that he may be obtaining some of his high raw numbers inefficiently.
If we take the most optimistic view of Westbrook, he’s right on that top ten bubble, but if we take the most pessimistic view, he’s pretty much a league average player. This nicely dovetails with the opinions from the original debate (and perhaps show which all-encompassing statistics the debaters found most credible).
Among traditional offensive metrics, Westbrook is at or near the top. Last year he scored 23.6 points per game, fifth in the league. He didn’t obtain those points very efficiently, however, with his true shooting % placing him 150th in the league, which even takes into account the fact that Westbrook gets to the line at the seventh best rate in the league and shoots free throws well.
Westbrook’s inefficiency is an especially large problem because he is teammates with Kevin Durant and James Harden. Last year Durant was the top scorer in the league, and Harden the 27th (in many fewer minutes than Durant and Westbrook). Durant’s true shooting % was 19th in the league, and Harden’s 6th, yet Westbrook only took 30 less shots than Durant and 637 (637!) more shots than James Harden. Fittingly, Westbrook’s offensive-rating (points scored per 100 possessions) was 131st best in the league, while Durant’s was 48th and Harden’s 10th.
I can’t overstate how important those efficiency differentials are. In any given game of basketball, there is a relatively finite supply of shots. Offenses spend the entire shot clock looking for the most efficient shot given the situation. Since practically every minute Westbrook is on the court either Durant or Harden is as well, he (and the rest of the offense) should be looking to get them the ball in good situations. Instead, Westbrook shot ten more times per game than Harden!
As far as his position as a point guard and distributor, Westbrook fares no better. He was tied for 14th in assists among point guards, but was 5th in turnovers. His Assist to Turnover ratio was 64th best among all of ESPN.com’s 73 qualifying players. I’m not a huge fan of such a strict definition of position (quick, what position does LeBron James play?), and thus give Westbrook somewhat of a break here.
Westbrook reminds me a lot of Allen Iverson, a player who got great numbers based on a high volume of minutes and shots, but wasn’t that efficient of a player. This is where advanced statistics are incredibly useful, letting us delve beyond box score numbers, where Westbrook looks great, to try an understand his true value. All the more damning is the fact that Westbrook is teammates with two of THE most efficient scorers in the game.
It is much more difficult to assess Westbrook as a defender because, in general, our understanding of how to measure defensive impact is light years behind offense. Here we generally have to rely on the smell test and box score numbers more than we do in evaluating offense.
The raw box score statistics favor Westbrook. Last year he was 6th in blocks among point guards and 5th in steals. He also fouled at a low rate, rarely giving opponents free throws. He is highly touted as being a good rebounder for a point guard and this grades out, as he grabbed the most rebounds per game among all point guards. He’s also second in rebounds per 48 minutes, so it’s not just a function of minutes played.
The advanced statistics don’t paint such a beautiful picture. Westbrook got the 50th most defensive win shares in the league last year, but much of this was a function of his high minutes, as his defensive rating was just 215th in the league. That defensive rating was 105, while Oklahoma City’s defensive rating was 103.2, discrediting the notion that Westbrook suffers because he was on a poor defensive team. Oklahoma City was the 10th best defensive team in the league, and on that team Westbrook’s defense was below average.
It is pretty clear that saying Russell Westbrook is the 9th best player in the league is quite an overreach; 30th seems to fit better, and I could be convinced that 50th is right. Westbrook clearly has the physical attributes to excel at all facets of the game like few other players, but basketball is a 5-on-5 game, not 1-on-1. The fact of the matter is that Westbrook is not elite at scoring, yet chooses to continue to shoot as if he is. If it were instead James Harden who shot ten more times a game than Westbrook, Oklahoma City would score 1.7 more points per game! Now, whether or not Harden would keep up the same efficiency is an open question, but it is certainly one worth trying to answer right?
As many of these statistical arguments go, a lot of the difference in perceptions can be traced back to the Yay! Points! Thesis. Scoring is how you win basketball games, and Westbrook scores a lot of points so he must be good right? Absent the context of team, efficiency, and defense sure, but if you’re going to throw that all out what’s the point?