The proliferation of advanced analytics has meant we now judge a scorer by volume and efficiency. It’s time to really add consistency to the mix.
The proliferation of basketball analytics has exponentially increased the number of ways in which we can describe, analyze, and understand the game of basketball. Advances over the last decade have taken us from simple per game stats into a world of more descriptive metrics like per 36 minutes, per 100 possessions, and beyond. But the fact of the matter is that one statistic still reigns king in the hearts and minds of fans: scoring.
And it should, at least in my opinion. Scoring points is still the ultimate objective of the game and might be the area where an individual player can have the greatest impact on winning. But while we have made significant advancements in how we think about scoring, through the use of per 36 minutes or per possession statistics, our discussions are still significantly limited by the fact that we look only at the average scoring for players, and not how it varies from game to game. Some players have become famous as streaky scorers, but our focus on consistency remains largely anecdotal, without a common metric to quantify it.
Let’s look at an example. Last season, Jamal Murray and Malcolm Brogdon scored at a very similar rate, putting up 20.1 and 19.7 points per 36 minutes, respectively. But the density plot below shows that Brogdon scored around 20 points per 36 minutes in a significantly higher percentage of his games than did Murray (excluding games in which either played under 12 minutes).
In other words, Murray had higher highs and lower lows, while Brogdon was much steadier in his performance. What if we had a metric to show that difference?
To create one, I borrowed a common approach from economics for measuring spread across variables with different means: the coefficient of variation. The coefficient is a simple measure that divides the standard deviation of a population by its mean. Since I am interested in the consistency of scoring, not necessarily consistency in the number of minutes played, I decided to calculate single-game per-36-minute scoring instead of just the total points in that game. I excluded any games in which a player played fewer than 12 minutes to cut out any wonky per 36-minute numbers that could result from a smaller sample. I then calculated the coefficient of variation for those per-36-minute scoring performances for each player in each season since 1980 (the first in which the 3-pointer was allowed). In its standard format, a lower coefficient of variation would imply that a player was more consistent in their scoring. But bigger usually feels better, so lets make it that way by inverting the coefficient of variation, meaning the higher your score the more consistent you are.
The final result: Inverted Coefficient of Variation of Scoring (ICoVS).
The plot below shows the relationship between players’ ICoVS and their per-36-minute scoring. Interestingly, higher scorers also tend to be more consistent on average.
So how well does this measure help us solve the problem we saw earlier with Jamal Murray and Malcolm Brogdon? Well, last season Brogdon put up a ICoVS of 3.25, the 25th highest mark in the league. Murray, on the other hand, scored an ICoVS of only 2.44, the 119th highest mark in the league. The difference we saw earlier in their consistency is now starkly apparent.
With this new approach for quantifying consistency, we can compare a significantly greater number of players without having to dig deep into their game-to-game differences or creating individual density plots. So who were the most consistent scorers last season?
For the most part, the list includes the usual suspects. On top of his dominant post-season run and second finals MVP, Kawhi can now also claim the title of the most consistent scorer during the 2018-19 regular season (as exciting as a ring, I’m sure). LeBron James came as a close second, with Harden following in third. One notable omission: Stephen Curry. Despite ranking fourth in per-36-minute scoring, Curry was significantly less consistent than his fellow elite scorers, coming 43rd in ICoVS. In fact, the same discrepancy was seen in his fellow splash brother Klay Thompson‘s numbers, who ranked 28th in points per 36 but 176th in ICoVS.
One explanation for this might be their higher reliance on 3-pointers. The graph below shows that there does exist a negative relationship between 3-point rate and ICoVS. In fact, last year James Harden was the first player ever to put up an ICoVS over 4 while taking more than half of their shots from 3.
But there were some exceptions, such as Buddy Hield, who ranked 19th in consistency despite hoisting 9 3-point attempts per 36 minutes last year.
A bigger question remains: who is the most consistent scorer ever? To answer this, I calculated career ICoVS for all players in the modern era, designated by the addition of the 3-point line in 1980. The most consistent scorer ever in that period? Kevin Durant.
Maybe that shouldn’t be a surprise, given that in August of 2017 Durant went on Bill Simmons’ podcast and said, “I’d rather have 30 for two weeks straight then have one big outburst.” But on a top ten list that contains LeBron, MJ, and Kareem, it’s still impressive that KD came out on top.
To see where all your favorite players ranked, check out the full database here.