Any Changes in Clutch Time?
The Celtics have been dramatically worse in clutch time in recent years than their normal minutes, is that true this year? Can it explain their recent skid?
As a disclaimer, this was written before the Rockets game on 01/21/2024.
Two nights ago, the Celtics dropped an incredibly close game to the Nuggets for their first home loss of the season. This drops their record in 2024 to 6-4, not only a much worse spread than their year average, but the only 10-game stretch of the season with such a record. Specifically, the Celtics have gone into clutch time 5 times in this stretch and are 2-3. While this is naturally setting off alarm bells for some Celtics fans with feelings that the Celtics cannot execute late in games, but these fears are largely unfounded.
For the Celtics, 21 games this season have reached what NBA.com considers “clutch time.” This is tied with 3 other teams for the 10th most such games in the league, a middling number. In these games, the Celtics are 13-8 but boast an 18.7 net rating in such situations. This is 80% of the team’s losses for the season and a win percentage drop-off of 14.3 points. For context, the last 5 championship teams have seen their winning percentages dip 5.2, 9.8, 17.5, 6.5, and 10.2 points each. The 2021 Bucks even had a losing record in clutch situations. While the Celtics' drop-off is on the high end of this range, this is more reflective of their outstanding win percentage in non-clutch situations than it is of serious crunch-time problems. In fact, the Celtics' 61.9% win rate in these close games would be the second-best of any of the aforementioned teams behind the 2020 Lakers at 66.7%. Additionally, not one of these teams won 60 games nor broke a 6.0 net rating. What we can draw from this suggests that the Celtics are actually winning their close games at a very high rate but appear to have trouble in those situations due to almost never being blown out.
This may feel like a cop-out to Celtics fans who feel that struggles in close games have been a calling card the last few seasons, but let’s compare. In 2023, the Celtics had the second-best clutch win % in the league but only the 8th-best net rating at 4.6. This is in large part because their offense completely collapsed. On the whole, the Celtics were the second-best offense in the NBA in 2022-23 with a rating of 117.3 and a net rating of 6.7. However, in clutch time their offensive rating dropped to 110.9. This season, their offensive rating actually increases from 120.4 to 124.0 in crunch time, contributing to their 8.9-point rise in net rating rather than the drop-off they saw last season. This trend can also be seen in 2021-22 when the Celtics’ 113.6 offensive rating plummeted all the way to 97.7, a bottom-5 clutch time offensive rating in the league. This will be tough to hear for Celtics fans, but the numbers are very clear - this team performs fundamentally different in the clutch than they have in the past two seasons.
This still doesn’t explain why the Celtics have received 40% of their losses in the last 10 games. This is a fairly small sample so it is difficult to try to draw conclusions In those games, 4 were comfortable blowout wins, 1 was a blowout loss, and 5 were close games that they lost at a higher rate than normal. Immediately we can treat the Bucks loss as an outlier that is not necessarily predictive. There is not a soul on the planet who believes the Bucks holding the Celtics to 102 points is going to be repeatable at scale or is representative of any broader trends in the Celtics season. The one part that loops into the rest of the losses would be an assessment that the Celtics are running out of legs. This is certainly something to watch for as the season goes on, but it is not clear from the data that this is any sort of trend. While their numbers are down slightly across the board during this stretch, this could equally be explained by fatigue as a minor cold streak.
Neither their shot profile nor their opponent's shot profile are seeing any sort of meaningful shift that could indicate more settling or less defensive movement. The Celtics have actually taken more wide-open 3s in their last 10 (19.4 per game) than they have for the season (18.3) and allowed fewer (21.4 to 19.8) from their opponents. They have taken fewer open 3s in that time, but it is also a margin of 1.1, and switching 1 open 3 a game to a wide-open 3 is not exactly a cause for concern. Their shooting and opponent’s shooting are roughly in line with season averages as well, even seeing slight bumps reflective of the slight change in shot profiles.
If the Milwaukee game is ignored as an outlier, the 3 losses are less substantial and can be treated as more of an aberration in clutch games, which are by their nature more fickle and subject to hot and cold streaks. However, it would be irresponsible to just assume variance without at least looking for plausible explanations. The simplest number that sticks out is their 2Pt% dropping from ~53% to ~40%. Interestingly, this is also paired with an increase from 53.7% of their two-pointers being unassisted to 62.5% being unassisted. The NBA website does not track shots by distance specifically in clutch time, but this could suggest a higher percentage of looks being midrange or forcing it through multiple defenders at the rim. It’s also worth noting that their FTs are slightly down from 3.8 to 3.4. This is not substantial and could absolutely be attributable to sample size, but could also suggest refs are allowing more contact or that the Celtics are settling for midranges instead of forcing the issue more often.
Jayson Tatum and Jaylen Brown specifically are taking more of their shots unassisted. While they make 87.5% and 58.3% of their clutch time 2s without an assist normally, those numbers are up to 100 and 66.7 in the last 10. This matches the eye test that we are seeing more isolation and midrange from them in those moments but it is still shocking that Tatum hasn’t been assisted on a single 2-pointer in clutch time during this span. Also noteworthy is that Jaylen has not hit a clutch time 3 at all during this span, full stop.
Part of what makes this analysis difficult is a lot of the Celtics' numbers are seeing disproportionate rises from two absolutely stellar performances against OKC and MIN. One could try to control for this by eliminating them from the set, but I generally consider it poor form to eliminate 40% of a dataset because it doesn’t fit with a general conclusion. This is going to lead to a rather unsatisfying conclusion this week, and that’s what we need to keep watching. It’s not really clear what is plaguing the Celtics. It’s possible to look at a series of outliers, for example eliminating Milwaukee and saying that the Indiana game should’ve been a win based on the last two-minute report and after eliminating both the sample is not even remotely concerning, but it’s also possible that a trend is developing. The ultimate problem is that there just isn’t enough data to find that trend and I’m not going to do a deep film dive just yet to see if I can identify it.
Anyway, thanks for reading! This week was mostly throwing a bunch of numbers at the wall and shrugging, we’ll be back next week with more!