12-21-2020, 08:33 AM
Wins vs. Expected Wins. Expect a (good) regression to the mean
https://www.theringer.com/2020/12/21/221...ns-nuggets
Dallas Mavericks
2019-20 record: 43-32
2019-20 Pythagorean record: 48.4-26.6 (-5.4 actual wins vs. expected)
Record in one-possession games: 2-11
The Mavericks were the NBA’s unluckiest playoff team last season: They landed the West’s no. 7 seed, but by point differential, they were closer to the Lakers and Clippers than any of the West’s other playoff teams.
That kind of extreme underperformance doesn’t matter for this season. Since the introduction of the shot clock, both the biggest overachievers (teams that outperform their Pythagorean expectation by at least 5 percentage points) and the biggest underachievers (teams that fall short by at least 5 percentage points) have regressed to the mean the following season, meaning their over- or underachievement doesn’t carry over.
Pythagorean Divergence a Year Later
Subset of Teams Year X Year X+1
Overachievers +6.6% +0.6%
Teams in the Middle -0.1% +0.0%
Underachievers -6.9% -1.2%
All teams in the shot clock era. Over- and underachievers had their record diverge from their Pythagorean expectation by at least 5 percentage points.
For instance, the largest overachiever in the 21st century is the record-setting 2015-16 Warriors, who won 73 games versus just 65 “expected” wins. After adding Kevin Durant, they had an even better underlying performance the following season—but fell to 67 wins, precisely their expectation by point differential.
On the other end, the largest underachiever this century (and the second-largest in history) was the 2011-12 76ers, who won 35 games in a shortened season, versus 43 “expected” wins. The following season, the 76ers actually overachieved, with 34 actual versus 31 Pythagorean wins.
Dallas’s greatest issue was its performance in close games. The offense that set league efficiency records ranked just 26th in clutch situations, and the Mavericks went 2-11 in games decided by three points or fewer—one of the worst marks ever for an otherwise winning team.
But record in close games is also a fluky stat that regresses to the mean over a large enough sample. There is nothing to suggest ineffective clutch teams in one season will remain ineffective the next. In the 3-point era, the correlation of record in one-possession games from one season to the next is just 0.1, on a scale in which 0 represents no relationship and 1 a perfect relationship.
Before last season, 32 teams in the 3-point era had won 20 percent or fewer of their one-possession games, like the Mavericks did in 2019-20. Yet that same group of teams won nearly half of its close games in the following season.
The Worst Teams in One-Possession Games, a Year Later
Season Wins Losses Percentage
Year X 43 254 14%
Year X+1 177 207 46%
Among 32 teams in the 3-point era that won 20 percent or less of their one-possession games
What does that history mean for Dallas? Well, if the Mavericks had gone even 5-8 in those close games instead of 2-11, they would have been the West’s no. 4 seed instead of finishing seventh and facing the Clippers in the first round. And that assessment comes even though Luka Doncic and Kristaps Porzingis missed time with injuries. So as the Mavericks enter the 2020-21 season, even with Porzingis still recovering from surgery, it makes more sense to think of them as one of the leaders of the West’s non-L.A. contingent than a typical no. 7 seed trying to inch higher.
It’s possible that the Mavericks are irrevocably ruined at the end of games—but it’s far more likely that in a small sample of scattered close games, they just had shots rim out at the wrong times. They didn’t have any notable clutch issues in 2018-19, when they finished 7-7 in one-possession games and rookie Doncic was one of the league’s best clutch players. Doncic’s clutch performance fell off in his sophomore season, with his clutch effective field goal percentage dropping from 51 to 40 percent. But the guy who made this shot in a playoff game will probably win more than two close games this season.
https://www.theringer.com/2020/12/21/221...ns-nuggets
Dallas Mavericks
2019-20 record: 43-32
2019-20 Pythagorean record: 48.4-26.6 (-5.4 actual wins vs. expected)
Record in one-possession games: 2-11
The Mavericks were the NBA’s unluckiest playoff team last season: They landed the West’s no. 7 seed, but by point differential, they were closer to the Lakers and Clippers than any of the West’s other playoff teams.
That kind of extreme underperformance doesn’t matter for this season. Since the introduction of the shot clock, both the biggest overachievers (teams that outperform their Pythagorean expectation by at least 5 percentage points) and the biggest underachievers (teams that fall short by at least 5 percentage points) have regressed to the mean the following season, meaning their over- or underachievement doesn’t carry over.
Pythagorean Divergence a Year Later
Subset of Teams Year X Year X+1
Overachievers +6.6% +0.6%
Teams in the Middle -0.1% +0.0%
Underachievers -6.9% -1.2%
All teams in the shot clock era. Over- and underachievers had their record diverge from their Pythagorean expectation by at least 5 percentage points.
For instance, the largest overachiever in the 21st century is the record-setting 2015-16 Warriors, who won 73 games versus just 65 “expected” wins. After adding Kevin Durant, they had an even better underlying performance the following season—but fell to 67 wins, precisely their expectation by point differential.
On the other end, the largest underachiever this century (and the second-largest in history) was the 2011-12 76ers, who won 35 games in a shortened season, versus 43 “expected” wins. The following season, the 76ers actually overachieved, with 34 actual versus 31 Pythagorean wins.
Dallas’s greatest issue was its performance in close games. The offense that set league efficiency records ranked just 26th in clutch situations, and the Mavericks went 2-11 in games decided by three points or fewer—one of the worst marks ever for an otherwise winning team.
But record in close games is also a fluky stat that regresses to the mean over a large enough sample. There is nothing to suggest ineffective clutch teams in one season will remain ineffective the next. In the 3-point era, the correlation of record in one-possession games from one season to the next is just 0.1, on a scale in which 0 represents no relationship and 1 a perfect relationship.
Before last season, 32 teams in the 3-point era had won 20 percent or fewer of their one-possession games, like the Mavericks did in 2019-20. Yet that same group of teams won nearly half of its close games in the following season.
The Worst Teams in One-Possession Games, a Year Later
Season Wins Losses Percentage
Year X 43 254 14%
Year X+1 177 207 46%
Among 32 teams in the 3-point era that won 20 percent or less of their one-possession games
What does that history mean for Dallas? Well, if the Mavericks had gone even 5-8 in those close games instead of 2-11, they would have been the West’s no. 4 seed instead of finishing seventh and facing the Clippers in the first round. And that assessment comes even though Luka Doncic and Kristaps Porzingis missed time with injuries. So as the Mavericks enter the 2020-21 season, even with Porzingis still recovering from surgery, it makes more sense to think of them as one of the leaders of the West’s non-L.A. contingent than a typical no. 7 seed trying to inch higher.
It’s possible that the Mavericks are irrevocably ruined at the end of games—but it’s far more likely that in a small sample of scattered close games, they just had shots rim out at the wrong times. They didn’t have any notable clutch issues in 2018-19, when they finished 7-7 in one-possession games and rookie Doncic was one of the league’s best clutch players. Doncic’s clutch performance fell off in his sophomore season, with his clutch effective field goal percentage dropping from 51 to 40 percent. But the guy who made this shot in a playoff game will probably win more than two close games this season.