A Graphical Analysis of Stephen Curry
And a Record-Setting Season of Dominance
Daniel Brink
A QUICK BACKGROUND
Since the onset of the 2015-2016 NBA season, the Golden State Warriors have been making basketball history by steamrolling their opponents en route to a 73-9 win/loss record, breaking the previous record of 72-10 set by Michael Jordan’s legendary 1996 Chicago Bulls team. The team has been operating at an offensive efficiency never seen before in the history of the league, boasts some of the best 3-Point shooting stats of all time, and has made it undeniably clear that - to the basketball enthusiast - a Warriors game deserves the designation of “can’t miss television.” While the Warriors have dazzled viewers by shooting 3-Pointers at 41.6% (best in the NBA) and garnering the best record in league history, fans of basketball, regardless of team affiliation, have the pleasure to bear witness to perhaps the greatest team performance of all time.
The team’s success alone is enough to make any casual fan excited when Golden State takes the court, but what is even more exciting is the fact that this historic run is occurring in conjunction with perhaps the most unbelievable individual season of all time. Reigning league MVP and Golden State Warrior point guard - Stephen Curry - is a national sensation for “breaking” basketball with his spectacular performances while running away with a second league MVP in as many years. So, how exactly is Stephen Curry “breaking” basketball?
In order to answer that question, we first have to look at what exactly it is that makes a team have success on the court. In basketball, and in sports in general, this variable is easily found - just take a peek at the team’s wins column in the standings.
The Golden State Warriors finished the season with 73 of these. Theoretically, if you could figure out exactly what individual player statistics contribute to tacking on another W to a team’s record, a general manager could construct a “perfect” team with players that fit these statistical molds to win every game - and I could display to you exactly how severely Stephen Curry’s play is affecting his team’s domination of the league in 2016.
Realistically though, it’s not that simple. A team is assigned a win as a whole, not a player, and it’s pretty difficult to pick out how just one player can garner a win for a team. Accordingly, we have to establish a “pseudo-variable” or sorts to replace wins: a counting statistic that can be recorded in splits for when a player is on and off a court. In order to do this, I downloaded each NBA team’s counting statistics over the past ten years, along with their total win counts via NBA.com, along with their total win counts in each season, and ran Pearson correlations on all of the variables to find out which variable was most highly correlated with winning.
I found that out of all the team’s counting statistics, the one yielding the highest correlation coefficient was Net Rating (denoted NetRtg), with a measure of .973. According to the NBA.com statistics glossary, Net Rating is a statistic that:
Measures a team’s point differential per 100 possessions. On player level this statistic is the team’s point differential per 100 possessions while he is on court.
Logically, it makes sense that this variable would be highly correlated with winning, as scoring more points than the other team results in a win, and with such a strong correlation coefficient and applicability to an individual player, it makes for the perfect substitute variable for wins.
Next, it was essential to find what player metrics were indicative of high NetRtg measures. To do this, I downloaded the past few years of NBA player data and ran a series of multiple linear regressions to try and establish which statistics were indicative of a high NetRtg, and by proxy, a high winning percentage. In my efforts, I regressed on metrics from individual players, five-man rotations, three-man rotations, and two-man tandems. However, finding a regression equation that appropriately fit was no simple task, and the r-squared value of these models only ever got as high as .340. If only effort was highly correlated with finding a successful regression equation…
Consistently, however, there was one variable that always established itself as an extremely statistically significant predictor of NetRtg - effective field goal percentage (denoted eFG%), which according to NBA.com, is defined as:
The statistic accounts for the fact that a three-point field goal is worth more than a two-point field goal, as described by basketball-reference. The formula is (FGM + 0.5 * 3PM) / FGA.
In fact, when I ran a stepwise regression model on NetRtg, in which the most significant predictor is inputted into the model first, it didn’t matter what other variables, interactions, or manipulations I tried to include; eFG% was always the first variable R chose to include in the model. Clearly, this means that eFG% is a very important metric for team success, and if nothing else, is a clear indicator of an efficient and effective basketball player.
So, how exactly is Stephen Curry “breaking” basketball?
Let’s start by taking a look at his eFG% …
ELITE EFFICIENCY FOR AN ELITE PLAYER
Of all qualifying players (minimum 58 games), Stephen Curry is dominating the NBA in eFG%, ranking second only to DeAndre Jordan, center for the Los Angeles Clippers.
But given the relative distances from the basket at which these players typically shoot the ball, it’s not at all farfetched to comprehend why Jordan’s eFG% is higher than Curry’s. In fact, it may have been wild to assume that a player with Curry’s long-range shot tendencies could ever eclipse a dunk-’em-all center like Jordan. That’s what makes Curry so special.
After Curry and Jordan, the top ten are rounded out by players with similar shooting tendencies to Jordan. In fact, Curry stands alone as the only guard in the group. These players all boast high percentages due to the relatively easy nature of the shots they take, but Curry, while bombing from long distance, still manages to keep pace with them.
To illustrate, below is an animated graph displaying each player’s eFG% from different distances away from the basket. The x-axis denotes each of 2016’s top 10 players in terms of eFG%, the y-axis determines each player’s eFG% at from that particular distance, and the size of the bubble determines the relative frequency at which each player takes that shot.
A few takeaways can be had from this graph. First and foremost, Curry is the only player in the top ten for eFG% that even attempts a 3-Point shot on a regular basis. Furthermore, at only 6’3" Curry is just as efficient, if not better, at short-range shots than his 7’0" counterparts. What is most astounding, however, is that his effeciency is higher from 30-34 feet away than most of the others’ are from a less-than-5-footer.
You as well might deem him a robot with that kind of nonhuman efficiency. Rarely, if ever, do you see a player shoot with such incredible effectiveness from all ranges, particularly so far away.
The man can hurt a team from 5 feet…
… just as easily as he can from 25 feet.
