I made a trip to Oracle Arena on December 19th to watch the Golden State Warriors take on the seasoned San Antonio Spurs. The previous two Spurs-Warriors matchups I had been to both went to overtime, so I was expecting a good game. What I did not expect was for the Spurs’ Big 3 (Tim Duncan, Manu Ginobili, and Tony Parker) to sit out and for the game to still go down to the final seconds. Neither did the rest of the country, especially when the shorthanded Spurs came away with the ‘W.’ Since this was a nationally televised game, the outcome reignited a spirited debate about whether the Warriors were a legitimate contender.
The Warriors have since gone 11-2 (most of these wins were on the road), beating marquee teams like the Miami Heat. This week, they are also top 3 in power rankings across the board. With some new moves on Wednesday that will strengthen the Warriors’ shallow bench, they are back in the mix.
However, I will stay away from using the loaded term ‘contender.’ After all, there are playoff teams and there are lottery teams. In order to have a shot at the title, teams need to minimize their weaknesses and maximize their strengths. At the end of the season, there is only one champion, so this discussion adds little value to understanding individual team performances. Though Golden State is clearly a force in the NBA (deservedly so), we should not ignore what happened on December 19th. Sitting in the stands after the loss, these were the thoughts that went through my head:
1. There are very few playmakers on this team.
2. Klay Thompson can be a one-dimensional player on the offensive end, but you really notice how bad it is when he is cold from beyond the arc.
3. What happened to Harrison Barnes and the Warriors’ second unit?
This got me really thinking, so I decided to delve deeper into the stats this week, focusing on Golden State’s offensive vulnerability…
20 is the Magic Number
Before addressing the aforementioned topics specifically, I wanted to get a sense for what objectively makes the Warriors “tick.” This is a fairly tall order, so I took a rough cut at this by looking at all 39 games Golden State has played so far (as of Wednesday morning), and I used a decision tree classification approach to determine which game stats were most indicative of determining whether a game was won or lost. The input variables I included in this analysis are all the basic box score statistics (assists, rebounds, steals, etc…) along with opponent win percentage.
Given the small sample size, I was not expecting too much to come out of this exercise. However, to my surprise, there was a single node that did a reasonably good job predicting the data. It is assists, and the magic number is 20. The model, in essence, says that if the Warriors have 20 or fewer assists in a game, then that is the strongest signal for a loss. Using this classifier alone, the Warriors predicted record for the season would be: 26-13. Their actual record is: 25-14. (For a more in depth discussion of the test and training data, click here.)
We know that the Warriors usually have pretty good ball movement based on the number of assists that Steph Curry racks up. In fact, they are the eighth best team in the league in terms of assists, averaging 23.2 a game. For context, roughly half of teams in the NBA average more than 20 assists per game. So what does this mean? Though this is not useful in predicting whether the Warriors will win or lose a given game, it adds a lot of descriptive insight with respect to their style of play.
When you actually examine the individual plays that lead to the Warriors’ assists, there is something really striking. When it comes down to it, a big reason why the Warriors are such a high assist team is because there are so few playmakers on the court…
Stay tuned for part two of this analysis, in which I delve deeper into the self-regulating cycle of assists, shot creation, and shot selection that define the offensive prowess and vulnerability of Splash Brothers & Co.