Reconciling Wins Produced with team chemistry
This week’s two-part Q and A with Stumbling on Wins‘ David Berri provided a perfect time to ponder how we evaluate the game of basketball.
From the looks of the comments section, many people on this site were introduced to Berri’s efficiency-driven Wins Produced stat for the first time this week. As you may have guessed, I do find great merit in Prof. Berri’s work and the advanced stats movement at large.
The biggest thing I like about Wins Produced is exactly what many of you commenters dislike about it — it defies conventional wisdom because it’s free of biases.
When you see Jason Kidd, Gerald Wallace and Marcus Camby in the top 10 and Kobe Bryant and Brandon Roy below the top 30, the first thought is that this system is insane. Of course, before jumping off the deep end it’s important to consider the fact that Kobe and B-Roy both suffered injuries throughout the year and this is a cumulative stat, but the beauty of this stat is that it can’t be swayed by SportsCenter.
No matter how glamorous a player’s game is, he can’t hide from Berri’s system if he’s not as efficient at producing wins as another player.
Berri himself took the time to defend this approach on his site in light of the attack from people on my site:
"As I noted back in 2007, all of the “advanced” statistical models (well, maybe not the Player Efficiency Rating) create rankings of players that defy conventional wisdom. That’s because conventional wisdom is driven by points scored. And the “advanced models” are not driven by scoring (well, except for the Player Efficiency Rating). So if you are an adherent to the conventional wisdom of the NBA, you are probably never going to like any of the advanced models.And that means fans of the NBA have a choice. They can simply follow the conventional wisdom. That means you look at a player’s scoring and believe that Allen Iverson, Carmelo Anthony, Rudy Gay, etc… are above average (if not absolutely great) players. Of course, when your team gets these players you may not see as many wins as you like. But then you can turn to the subject of chemistry (or fairy dust, astrology, etc…) and still remain pretty happy. …So the advanced models can provide a better way to look at basketball (for whatever that’s worth). But they do come with a cost (i.e. you have to do some thinking to appreciate how these models work). And this is why I think we are unlikely to see a day when most NBA fans abandon the “conventional” wisdom."
Seth Pollack over at Bright Side attacked Berri’s explanation from Part 1 that teams that exceed expectations enjoy good chemistry but teams that fail to meet expectations suffer from bad chemistry:
"Berri’s problem is the lack of definition, which really is just another way of saying he lacks information, imagination, people skills and interest.Just because “team chemistry” can’t be counted doesn’t mean it doesn’t exist and it doesn’t mean there’s not a serious and rigorous approach to building it.For Berri and his stat-loving friends to discount something they don’t understand is just as ridiculous as me ignoring the value of the Moneyball crowd because my math skills are below fifth grade level (which they are not)."
At the risk of taking the easy way out, I kind of agree with both Berri and Pollack. The Wins Produced stat explains how basketball games are won free of bias based on factors that have been proven to lead to winning.
However, I do believe there are some intangibles that cannot be picked up in any box score or even a statistic as sophisticated as Wins Produced.
One player who immediately comes to mind is Channing Frye. Wins Produced hates Frye, as he only produced 1.3 wins last season. But Suns head coach Alvin Gentry once compared Frye to an elite wide receiver who gets double covered and thus opens the game up for the rest of his teammates. Did Frye’s ability to space the floor show up in the form of one of his teammate’s Wins Produced output? Perhaps.
Then there’s Steve Nash. Berri wrote in Part 1 that Nash improving by 3.4 wins between 2008-09 and 2009-10 is one of the biggest reasons the Suns did so much better than expected. My theory would be that shedding Shaq and playing with a group of teammates he truly trusted in a system that suited him (team chemistry) led Nash to produce more wins.
I believe very strongly in both team chemistry and its positive externalities as well as in advanced stats like Wins Produced. Perhaps the fit of a team can make a player more productive by allowing him to create more efficient scoring opportunities for his team. The same could be said for a player’s defensive fit.
In this sense I think teams should draw upon both realms when putting together a ball club and doling out minutes. I also feel like adjusted +/- data should be considered, especially as it relates to determining which lineups play best together.
We are in a day and age in which sophisticated analysis can give us a deeper look at the game than the most experienced scout could ever decipher with only his two eyes. We should not be afraid to buck conventional wisdom in this regard, but at the same time a lineup of the top producers of wins at their position might not make for the most efficient team in actuality due to usage and the law of diminishing returns.
It’s still important to find pieces that complement each other and thus become more efficient at creating wins when paired together, and that’s why it’s vital for NBA front offices to consider both advanced analytics as well as less defined concepts such as chemistry when putting their basketball teams together.