Shooting Quality and Stephen Curry's Slump

    It’s easy to forget how well Stephen Curry started the season. His MVP candidacy, and for a long period he was full blown frontrunner, was built on more than the early dominance for the Warriors in terms of record. After 22 team games, Curry was averaging nearly 28 points, shooting almost 42% from deep on 13 attempts a game and playing the best defense of his career. 

    December, however, saw a precipitous drop in Curry’s efficiency – even if the Warriors have through their elite defense managed to keep the train running along. The case he was building for a third MVP award feels already lost in the collective memory as he, to put it bluntly, bricked his way out it. In the following 16 team games (out of which Curry played 14), he’s scoring 25.6 points per game on 34.6% from deep and an extremely more concerning 37.8% from the field overall. Even the free throw shooting is below 90%, pointing towards a slump of the kinds the best shooter in NBA history has never seen before.

    But despite his positive in the offense still being a clear positive and despite a legitimate discussion that should exist around what is happening to Curry’s ability to finish at the rim, I bring up the difficulties from deep as only an excuse to look at the evolution of a Shot Quality metric since early December. But first, I need to briefly explain how I computed this metric, where I even got it from and what it intends to capture.

CraftedNBA (I cannot praise this website enough), presents a metric called Shooting Quality, or SQ, that is formalised as such:

((Spacing x 2) + ((Standardized Padded FT% + Standardized Padded 3P%) x 5)) / 7

    The immediate take away is that it values a spacing metric with a weight of 2/7 and gives a larger weight of 5/7 to a combination of adjusted shooting percentages and before trying to recreate this metric, we first need to unpack these components. Starting with the Standardize Padded FT and 3P percentages, we turn to an article written on Kostya Medvedovsky’s blog discussing the “padding” of a player’s performance with an adjustment league-wide average and the use of a value that would act as a stabilization rate, as to mitigate noisy-small samples. Using the formulas and padding values presented in his article (242.61 3-FGA and 24.11 FTA for 3P% and FT%, respectively), we arrive at the following:

Standardized Padded 3P% = (3PM + 242.61 x LeagueAverage3P%) / (3PA + 242.1)

Standardized Padded FT% = (FTM + 24.11 x LeagueAverageFT%) / (FPA + 24.11)

    Turning now to the Spacing component, CraftedNBA themselves have defined it as taking into account “both attempts as well as accuracy, with a slight adjustment towards attempts”. 

Spacing = (3PA * (3P% * 1.5)) - LeagueAverageEFG%

    Spacing as a concept in the NBA is nebulous - there is a clear know-it-when-you-see-it aspect to it. But it is clear the ability of a player to “space out” and distort an opposition defense with his shooting is conditional on a threat perception that requires as much a volume of attempts as well as efficiency. It is more valuable to an offense to have a player shooting 39% on 5 attempts per game than 45% in 1, as well as having an increase in volume that comes from shot versatility (such as being able to pull up, shoot in movement, step-back, etc…). While we are at the moment not accounting for this “shooting vocabulary”, it is somewhat mitigated by a survivor bias that would mean that the best volume shooters only have said volume because they have multiple ways to generate efficient shots. This is the point I mention that CraftedNBA also blends this Shooting Quality formula with the career spacing rating. I did not do this so, for that and other reasons, it’s fair to expect I will have different - and I’d wager on ‘worse’ - results.

 

    Regardless, with all the pieces being in place, and using stats from Basketball Reference, we are now able obtain our Shooting Quality metric and take conclusions. The following gif shows the outcomes generated in the dates of December 13th, January 4th and January 8th:

    Because the metric is normalized to the highest value (so whoever leads will always have exactly SQ = 1), it is a good way to see how much Stephen Curry has been dragged down towards the rest of the distribution – closer to the elite “mortals” of 3 point shooting when in early December, Curry might as well be in a different graph of his own.

    Let’s focus on the landscape as of the day of writing this article and, at the risk of being simplistic because the metric including a free throws component, acknowledge some other protagonists of SQ and the many three pointers they have produced:

    • Patty Mills (42% on 7.6 attempts per game) and Buddy Hield (38% on 9.5 attempts per game) have been leading the non-Steph sweepstakes the entire season.

    • ‘No-brainer-all-Star’ Fred VanVleet has been solidifying his case of late with terrific performances and his shooting is now at 40.5% on 9.1 attempts per game.

    • Desmond Bane has developed a devastating on-ball game in his second year but it wasn’t at the cost of his reputation as a deadeye shooter, currently making 41.6% on 6.9 attempts per game, nearly 3 more than last season.

    • It’s a very simplistic take but players like Kevin Love (42.6% on 6.3 attempts per game), Cameron Johnson (44.4% on 5.7 attempts per game) or Eric Gordon (44.8% on 5.1 attempts per game) – meaning, players that are well above the trend of shooting quality vs volume – might as well just start bombing 7 or 8 times a game.

    Before concluding, it is important to remind ourselves that Stephen Curry is notorious for having his best run every year post-All Star Break, like clockwork. If this is to hold in 2022, the slump might also quickly leave the ever-changing discourse of the basketball world. 

    I should also note that only 27 players have shot at least 243 attempts this season, meaning that almost all players have not crossed the line of the padding value and still have the stabilization rate as the biggest component of their metric. There is still a long way to go and it is very likely that the overall distribution of SQ will look much different at the end of the year as it does today. Whichever way it goes, I will, with a nerdy sense of glee, continue to monitor it and its mission to capture the level of destruction of the long range bombers in the NBA.