clock menu more-arrow no yes

Filed under:

No Mo: Why End-of-Season “Momentum” Won’t Help (or Hurt) in the Playoffs

The Houston Rockets are the hottest team in the NBA coming into the 2019 postseason. Unfortunately, it doesn’t matter. Let us explain.

Getty Images/Ringer illustration

The Rockets enter the playoffs with momentum. The Nuggets do not. The Jazz do. The 76ers do not. The Thunder do, or maybe they don’t. The same dilemma afflicts the Celtics. The Magic might have the most momentum of any team in the league.

Despite copious research showing that momentum in sports is a myth, at least on a large scale, fans and analysts can’t help but discuss the concept this time of year. It’s still tempting to preview a playoff series by noting that, say, the Rockets defense looks mighty formidable lately, and the Nuggets have had trouble scoring, and the Sixers haven’t jelled as expected since February’s trade deadline.

Here, for reference, is a chart that shows how hot and cold the 16 playoff teams are by comparing their winning percentage in their last 20 games with their overall season winning percentage. (That context is included because a 14-6 record from the Magic is more impressive than the same record from the Warriors.)

2019 Playoff Teams by Relative Momentum

Team Last 20 Games Winning % Full Season Winning % Difference
Team Last 20 Games Winning % Full Season Winning % Difference
ORL 70 51 19
SAS 75 59 16
HOU 80 65 15
LAC 70 59 11
POR 75 65 10
UTA 70 61 9
GSW 70 70 0
BOS 60 60 0
DET 50 50 0
BRK 50 51 -1
OKC 55 60 -5
TOR 65 71 -6
DEN 60 66 -6
PHI 55 62 -7
MIL 60 73 -13
IND 40 59 -19

And now forget everything that chart taught, because none of it matters. Trying to predict the future from end-of-season records is the equivalent of sports haruspicy. Based on NBA playoff history, there’s no reason to expect any of this purported momentum, positive or negative, to carry over to the postseason.

Last month, NBC Sports’ Tom Haberstroh showed that how the top-two seeds in each conference perform down the stretch doesn’t matter for the playoffs. And earlier this week, analyst Dean Oliver looked at regular-season results since 2005 on a month-by-month level and found that records from March and April are no more predictive—and if anything, less predictive—than records from earlier months.

We can expand that kind of analysis further by increasing the sample size and looking at historical playoff results in a few ways. Let’s start simple. First, the larger the regular-season sample, the more predictive it is in forecasting which team will win a given postseason series. The following numbers come from looking at playoff results since the field expanded to 16 teams in 1984. (The lockout-shortened 1999 and 2012 seasons were cut from this analysis.)

  • Teams with a better regular-season record won 78 percent of playoff series.
  • Teams with a better record in the second half of the regular season—the last 41 games—won 74 percent of playoff series.
  • Teams with a better record in the last 20 games of the regular season won 66 percent of playoff series.
  • Teams with a better record in the last 10 games of the regular season won 61 percent of playoff series.

That’s a good starting point, and it refutes the possibility that late-season record is a better indicator than overall record. But few people would assume that’s the case—nobody would take an 8-seed over a 1-seed just because of a disparate final stretch of the regular season. These numbers are also skewed because the team with a better overall record is probably the team with the better 10- and 20-game records, too. What about when the two don’t match?

The results don’t suggest any real evidence for momentum there, either. Overall, teams with a worse regular-season record sprung the upset in 22 percent of series. When the lower-seeded team ended the regular season hotter, the upset rate barely budged. Lower-seeded teams with a better record over the last 20 games completed just 24 percent of upset opportunities; with a better record over the last 10 games, they completed just 26 percent.

It’s also hard to find any individual examples of teams that converted late-season momentum into a lengthy playoff stay. Of the 20 teams that most overperformed their overall record over their last 20 games of the regular season—in other words, the 20 hottest teams relative to their established level—not one reached even the conference finals. Yet of the 20 teams that most underperformed down the stretch—the 20 coldest teams—eight reached the conference finals, and five went on to the Finals.

Care for some specific recent examples? Here’s one from last season and one from five years before that.

  • Last year’s 76ers went 18-2 in their last 20 regular-season games and 10-0 in their last 10. Only one other team since 1984 has entered the playoffs with at least 10 consecutive wins. All that momentum led to an unsatisfying second-round loss after Brad Stevens’s Celtics exposed Ben Simmons’s greatest weakness. (The other 10-0 team, the 2004 Spurs, also lost in the second round because apparently momentum is no match for magic.)
  • The 2012-13 Nuggets entered the playoffs on a 17-3 stretch, including 8-2 in their last 10. They lost in an upset in Round 1 in the first of many triumphs to come for Steph Curry and the Warriors.

Polar-opposite specifics can be found among the cold teams:

  • Last year’s Warriors went 10-10 in their last 20 regular-season games and 4-6 in their last 10. They won the title.
  • The 2012-13 Spurs went 10-10 in their last 20 and 3-7 in their last 10. They reached the Finals and would have won the title, if not for Ray Allen’s momentous 3.

So when forecasting upcoming series, history says overwhelmingly to focus not on a team’s recent form, but rather on its overall body of work. Injuries matter, of course, so Indiana’s poor late-season performance without star guard Victor Oladipo shouldn’t be discarded. But in general, trying to draw conclusions about the future from a limited slice of the very recent past is a fool’s errand.

In analyzing overall team playoff results since 1984, the R-squared value of regular-season winning percentage is 0.45. In less nerdy terms, that means regular-season winning percentage alone—so not factoring in roster strength, or injuries, or opponents, or home-court advantage—explains 45 percent of a team’s playoff success. That’s a meaningful indicator; it’s almost half!

The smaller slices of the season fare far worse in their predictive power. The final 10 games of the regular season explain only 6 percent of a team’s playoff success, which might as well be nothing at all. (This table shows the R-squared for all playoff games won; the results are essentially identical when looking at playoff series won instead. They’re also essentially identical when looking at point differential instead of winning percentage.)

Larger Samples Are Better Samples

Season Span Playoff Predictive Power
Season Span Playoff Predictive Power
All 82 Games 45%
Last 41 Games 31%
Last 20 Games 16%
Last 10 Games 6%

Why is this relationship so weak? There are plenty of possible reasons—from an uptick in missed games down the stretch for star players on top teams (like with Giannis Antetokounmpo this year) to inconsistent scheduling in the final weeks (are the Jazz really scorching, or did they just enjoy an end-of-season cakewalk?) to matchups being far and away the most important factor in any given playoff series.

That doesn’t mean to never pick a hot team, or pick against a cold team, in the playoffs. Want to pick San Antonio to upset Denver, or Brooklyn to upset Philadelphia, or Houston to advance past Golden State to the Finals? Go ahead. But focus more on the matchups and teams’ specific strengths and weaknesses. Misconceptions about momentum won’t help.