During my top-40 player series, there were a number of comments about the difference between regular season and playoff impact among all-time players. When calculating championship odds, do we underrate superstars by focusing on their regular season impact metrics? Is there evidence that stars are more even valuable when the chips are on the line?
The obvious way to answer this involves an examination of playoff-only adjusted plus-minus (APM) data. But since plus-minus metrics are noisy, we’d need multiple seasons to offset variance from the small samples. As far as I know, there isn’t a robust, publicly available APM across multiple postseasons. (Most APM studies incorporate the playoffs into the entire sample, sometimes giving it additional weight.) So instead, I turned to augmented plus-minus (AuPM), which blends the box score with unadjusted plus-minus data to spit out a pretty accurate estimate of APM.1
Since we’re talking about CORP values here, let’s compare AuPM on a per game basis. So a +6 AuPM player who plays half of the time (24 minutes per game) yields a per game AuPM of +3. Using this approach, the best regular-season AuPM mark on record (available league-wide since 1994) came from LeBron James in 2009, who was worth about 7 points per game. Since postseason samples are so small, I examined three-year stretches with a minimum of 1,000 minutes played (still a noisy cutoff) to see if anyone topped 7 per game in the playoffs. Nine players eclipsed 5 points per game, and two went over 7, including LeBron himself:
Tim Duncan is the pace-setter here, with his blistering combination of box-score dominance (10.3 Box Plus-Minus) and on-off impact (+26.8 net) in nearly 2,000 playoff minutes during this stretch. Of course, the off samples in these cases are almost always too small (335 minutes for Duncan), so it’s difficult to completely eliminate noise. Nonetheless, two results of moderate samples outpace the regular-season leaders, and a handful of others cluster behind the best non-LeBron seasons on record. Through this lens, elite regular-season and postseason impact looks quite similar.
I also ran four-year values to collect larger samples. In such cases, only LeBron was over 5 points per game (with an AuPm of +6.9 from 2014 to 2017). Heck, Draymond Green, Russell Westbrook, Manu Ginobili and Shaquille O’Neal were the only others to top 4 points per game across four-year stretches. Longer periods like this allow for more stable off-court samples, but they begin to smooth out peak performance that might only last for a year or two.
One final piece to consider: In order to amass a decent sample, players need to advance deep into the playoffs, which means the competition becomes harder. High seeds reaching the third or final round will typically face a schedule that is 3 or 4 points tougher than a regular season slate. I’m not sure there’s a perfect way to account for that using this method, but the ability to hold similar value is generally more impressive in this context.2
Given the increase in schedule strength, it’s possible these high-end seasons are a hair more valuable than the full-season number suggest. Otherwise, it appears that top-end value for superstars isn’t materially greater in the playoffs than in the regular season, and many superstars fail to match their regular season value.
- For this research, I used per 48 minutes as a stand-in for per 100 possession data. Data was collected from nba.com, which has playoff plus-minus since 1997.
- It would be less impressive if teams were suddenly destroyed by the superior competition. During Duncan’s ’03 stretch and LeBron’s ’10 stretch, their teams were about +7 per 48 minutes with them on the court, about 3 points worse than their regular season numbers.