Jayson Tatum's Shot Profile: Chucking Wins Games

Following up on a post from last year, we revisit a full season of Tatum's shot data and assess what we think works

There has been a lot of noise in the early season about Tatum’s shot selection.  He is up to almost 8 pull-up 3-point attempts per game, nearly 2 higher than his average last season.  A lot of people have said this is the result of “settling” or falling in love with his jumper and that his shot selection is undesirable, but I’d like to challenge this assumption.  Last year, I wrote a piece about being willing to sacrifice consistency for “auto wins.”  I now want to revisit some of that data and talk more about what Tatum’s shooting means for the team as a whole.  I’ll note here that I care very little for how Tatum’s shot profile affects his individual stats, only how it appears to impact the team’s results.  For that reason, everything will be couched in relation to winning.

A quick note on methodology.  All data was collected from the NBA’s official website.  It includes all games Jayson Tatum played in with the exception of the game on 03/09/2024 against the Suns since the NBA lacks tracking data from that game and therefore no commentary can be made on how his pull-up shooting affected that game.  This leaves 73 regular season and 19 playoff games to be assessed.

Jayson Tatum took 5.7 pull-up 3-pointers per game in the regular season, 5 per game in the playoffs, for 5.6 per game across all 92 games.  He hit these at clips of 34.9%, 27.4%, and 33.5% respectively.  Naturally, this was the subject of a lot of conversation.  Tatum is 6’10, 210 lbs, and shot 71% in the restricted area.  Despite 3 being much larger than 2, we are looking at a difference of nearly 40 points per 100 attempts in favor of the restricted area looks.  Even if we expand to all of Tatum’s three-point attempts, and then subsequently limit to only his wide-open (defenders more than 6 feet away) looks, restricted area attempts still generate nearly 10 more points per 100 possessions.  This is without even factoring in free throws, which tend to favor getting closer to the basket and being more physical.  This poses the question: why not go inside instead of settling so much?

The answer has two main components.  The first is an argument from risk.  While a restricted area attempt of any sort is better than even a wide-open 3, players cannot simply choose to get into the restricted area.  Many attempts to do so will end up as a pass, a turnover, or settling for a slightly further out shot.  While Tatum is fairly effective at keeping his turnover count low and is an excellent passer, part of this is in his style of play.  By not forcing the issue in places that will be more difficult to get himself out of, he limits the risk of explosive plays going in the other direction.  In terms of those lesser shots, Tatum is not exceptional.  He shot 36.9% and 38.9% on non-restricted area paint attempts and midrange attempts respectively last year.  Neither of these numbers are not much to write home about and are in fact relatively poor.  Needless to say, these are not Tatum’s best spots on the floor.  He also shot roughly as many of those combined as he did restricted area attempts last season, suggesting that on possessions he is shooting inside, he only gets to the rim about 50% of the time and has to settle for shots that are nearly half as effective as his pull-up three the other half.  The broad point here is that the threes have a much lower risk threshold and a more consistent outcome.  You may note this cuts against the previous point about sacrificing consistency for explosions, but that brings us to the next point.

The team performs better when Tatum is drawing attention on the perimeter than on the inside.  One very interesting note from last season is that Tatum’s restricted area attempts had a slight negative correlation with wins (r = -0.15).  This is likely the result of his inside attempts going up when the 3 isn’t falling (r = -0.19 with pull-up 3 % and r = 0.23 with pull-up 3 misses) rather than going inside being bad for the team, but it does suggest that the 3s going in at a high rate tends to be more beneficial than the RA attempts which compensate for them.  It also bears noting that both pull-up 3 attempts and pull-up 3 misses have a higher correlation with RA makes than they do RA attempts - suggesting that even just pulling defenders out gets Tatum himself easier looks at the rim regardless of if they’re falling.  We don’t see the same correlation with his pull-up 3 makes, but that’s likely because when his jumper is falling he doesn’t go inside as much since the score explodes anyways and there is no reason to put the extra strain on his body.  

The most basic number I want to highlight is the correlation between wins and pull-up 3 attempts we saw last season (r = 0.22).  Across all the data I analyzed, only a small handful of things Tatum did had stronger correlations with winning: hitting at least 2 pull-up 3s (r = 0.27), total made wide-open 3s (r = 0.32), and total 3-point volume (r = 0.25).  Notable here is that volume on pull-ups outperformed both volume of makes and percentage on the same shots, all driving-related stats (including drives, points off drives, passes off drives, and assists off drives), and anything that involved a 2-point shot (restricted area attempts, anything in the paint, and mid-ranges).  Of the 3 variables that outperformed it, two of them (hitting at least two and made wide-open) deal with stats that are dealing with low values (hitting 2 being a boolean reaching 1 at the highest and wide-open shots being a very infrequent occurrence for Tatum, averaging roughly 1 attempt per game) that tend to skew correlations and make them less reliable.  The third variable that outperformed it is a broader category that contains in whole of the smaller category to which we are comparing.  Some might ask about the playoffs when his shot wasn’t falling at all, but the numbers there are even more shocking.  The correlation between pull-up 3 volume and wins was even higher (r = 0.34) and was only outperformed by wide-open 3-point volume, total 3-point volume, mid-range volume, and mid-range makes (volume, not percentage).  Of course, this sample is small and with only three losses we can expect some weirdness in the data set, but it is noteworthy that the volume of 3s was largely still king even when shooting a terrible 27% on them.

This is a long and wordy way of saying the following: there is reason to believe that no single thing Tatum does on the basketball court impacts winning as much as simply jacking up as many 3s as he gets a chance to.  Explaining this would require its own deep dive but I think the obvious answers (gives the Celtics chances to get lucky on his makes while also opening up the paint and the corners by forcing stronger coverages above the arc) make the most sense.  Regardless, the impact appears clear: the volume wins games on its own.

This claim may sound absurd so I’ll add some qualifiers, but I will note it largely holds up in the face of more data.  If we limit the data to only games Tatum took above his average pull-up 3s (6 or more) we suddenly get a negative correlation between attempts and wins.  This suggests that there is an upper bound on the efficacy of this strategy.  It is worth noting, however, that we don’t appear to be super close to that threshold.  Within the above-average attempts set, the Celtics won 86.4% of their games, a 71-win pace over 82 games.  Even limiting the set to 8 or more attempts (above his current average for the 2024-25 season), the Celtics win 82% of their games, a 67-win pace.  The averages for these sets are 7.3 and 8.6 respectively.  While we start seeing a downturn somewhere, both of these numbers are better than how the Celtics performed on average last year, suggesting that his current 7.5 attempts are closer to the optimal number than last year’s 5.6.  I’ll also note here that I ran a chi square calculation on the above average attempt outcomes and below average and it returned a p value of 0.08, or about an 8% chance that these differences are attributable to variance. While this wouldn’t pass scientific muster (usually 5%), it should be persuasive for most of us to believe.

The other major factor to consider is that in last year’s data set, there was almost no correlation between volume and percentage (r = 0.04).  To simplify, he had roughly the same percentages in games he took 2 attempts as in games he took 8 attempts.  There are a number of reasons this could be the case, but the one that makes the most intuitive sense to me is that Tatum’s feel for what constitutes a “good” shot relative to what is actually a good shot does not fluctuate much game to game and his volume is dictated more by how many times he gets a legitimately good look than what he is looking to force.  This would suggest it is not so simple as “get more shots up to win more” but rather “when teams guard him in such a way he feels comfortable getting more shots up, they win more.”

This second qualifier is much more significant to me than the first.  As noted, the first qualifier suggests there is an optimal number of looks for him is probably somewhere above 5.6 but below 8.6 and closer to 7.3 than to either 8.6 or 5.6.  This second qualifier suggests that if the volume is going up, it could signal one of three things: 1. Teams are guarding him looser on the perimeter 2. He has improved his shooting and feels comfortable with a wider range of shots or 3. He is forcing weaker looks in an attempt to increase his volume.  The first two would both simply be positives for the Celtics while I think the third is the scenario many Celtics fans are currently anxious about.  

We don’t have a lot of data so far.  As of this piece being written, the Celtics are 9-3 which means any correlative data is going to be a bit wonky due to the extremely low loss count and also just a fairly small sample, so take anything with a grain of salt.  That said, early returns are good.  Pull-up 3 attempts have nearly double the correlation coefficient with wins of pull-up 3 makes and have extremely strong correlations with his restricted area attempts as well.  I do want to heavily cabin this data by noting there is a substantial downward trend in attempts recently.  One could say this is teams reacting to his hot start and trapping him deeper, but this would also help illustrate the Celtics’ recent struggles.  With Tatum unable to get these looks off, the duty will fall on other players to react to the traps and pressure defenses.  With Jaylen Brown and Kristaps Porzingis out, the Celtics have struggled with this.  

I’m not going to say that the Celtics' offense is an issue - quite frankly the team is on a 61-win pace and has a +9.8 net rating - it’s fine, but I will say that any attribution of possible struggles to Tatum’s shot profile appears misplaced on his pull-up 3 attempts.  The data continues to bear out: when Tatum is able to get up pull-up 3s at high volume, the odds of the Celtics winning the game go up.