Analytical Mailbag: HUSTLE
No, not the movie. High-effort Underlying Statistical Tracking of Live Energy.
Hey TZM,
Which players hustle?
Before diving into the numbers, it’s important to clarify what hustle means. Hustle is one of the more elusive elements of the game to quantify, often leading to misplaced takes about certain players. Some analysts use physicality as a guiding metric, a player’s willingness to commit hard fouls or embrace contact might lead to some praise of them as a player who “hustles”. A more nuanced basketball philosophy points out the flaw in this notion. A physical player could hustle, but it isn’t a direct correlation.
While it's easy to point out what hustle isn't, figuring out what it truly is has proven to be much more difficult. I finally settled on something I thought made a bit more sense: hustle is a willingness to do the right things consistently, regardless of outcome. This is what makes “hard fouls” not a great feature in hustle. That type of physicality doesn’t necessarily indicate any positive outcome over a bunch of repetition.
For the purpose of your question, I decided that it would be clearer if I focused on defensive hustle. I decided to settle on this because if you’re looking for “gritty” players, it’s important to know which side of the ball they tend to focus on. This is an important distinction that helps for a clearer purpose when defining the stat.
Methodology:
After reviewing the NBA's advanced hustle stats, I selected the following features as key indicators of hustle:
Deflections: I considered choosing steals instead because of an 82games.com article by Roland Beech (now with the Kings) which showed that non-steal deflections actually didn’t correlate much with winning. However, I do think that all deflections, positive outcome or not, do indicate hustle and a willingness to keep active hands.
DEF Loose Balls Recovered: Players chasing down loose balls have clear impact on the game, granting new possession - potentially a four-six point swing.
Charges Drawn: Charges show a player’s willingness to put their body on the line instead of giving up 2 points.
Contested 2PT Shots: Contesting two pointers shows attentiveness on defense. Did not consider blocks because players that chase blocks instead of staying in the right spot shouldn’t be rewarded.
Contested 3PT Shots: Closeouts are a valuable part of hustle.
DEF Box Outs: Defensive box outs, instead of defensive rebounds, show consistent effort on the defensive end - and has good correlation with team rebounding.
I have seen some people attempt to calculate a similar number in the past, and oftentimes, they just choose the stats, sum them up, and call that hustle. But without correct normalization, it often skews towards big men, who contest more shots and box out more often. Accounting for positions allows for a clearer idea of effort.
After some deliberation, this is how I decided to calculate HUSTLE(High-effort Underlying Statistical Tracking of Live Energy).1
All stats are calculated per 36 minutes, so that lower usage players aren’t penalized. It goes against the spirit of hustle to show bias in favor of more talented players! The more important distinctions can be seen in the Z-score calculation. The mean is calculated by position. Alex Caruso not contesting as many two point shots as Brook Lopez doesn’t suggest a lack of hustle - rather, it’s the scheme set in place.
Before weighting the stats, I decided to take a peek at what the raw results look like. The usual suspects all show up. Without the weights, Draymond's strong defense on both 2PT and 3PT shots pushes him to the top. This isn't to suggest that Draymond lacks hustle, but it highlights the need to weigh 2PT and 3PT a little less than other stats.
The weights, which are hidden for now, were first calculated through regression analysis against DBPM, which is a flawed stat when it comes to making opinions about players, but important to see impact over minutes. These weights were adjusted slightly based on other correlation tests. Here is the distribution for the weighted HUSTLE, accompanied by a box and whisker plot to illustrate the variance in HUSTLE across all positions, showing relatively consistent medians with some variation in spread, particularly for power forwards.
Results:
The results were interesting, with TJ McConnell and Alex Caruso leading the way at the top. McConnell has earned a reputation as a pesky defender in recent years, and his play earned him some notability in the playoffs last season. Caruso, the winner of the 2024 NBA Hustle Award was unsurprisingly near the top of the list. Overall, the list seems mostly sound - Donovan Mitchell and Kyrie Irving also find their way on the list despite having reputations as being cones on defense. Perhaps their limitations are primarily due to their size rather than effort.
Conversely, the bottom 20 players contain some surprises too. I actually thought Sam Hauser was one of the more active defenders in the first few games of the Finals this season. Perhaps I should do a further study on elevation in the playoffs.
Some of the NBA's household names show a wide distribution in their results. It’s possible that high HUSTLE scores, as seen in players like Joel Embiid, Anthony Davis, and Zion Williamson, combined with fragile bodies, could be linked to their ongoing injury concerns. On the other hand, Jayson Tatum, often regarded as a good defender, shows a weaker signal of activity.
Final:
I do think it’s fair to say that the names you see at the top of the list are some of the more active players on the court - but I think it would be disingenuous to consider this stat as anything more than an indicator of activity. To truly understand hustle, more nuanced stats and a refined weighting system would be essential.
What am I listening to?
Maybe I should’ve saved this one for a Steph Curry article. Still though, John Mayer!
My main motivation to do this type of work is to come up with bad acronyms. Just wait till I come up with BOLBOL(Basically Overwhelming Length, But Offensively Lacking).