This afternoon, I'm heading down to Portland for my last live NBA game of the 2011-12 season, and that's got me thinking back on the season opener, when the Trail Blazers hosted the Philadelphia 76ers. Throughout the season, the Blazers and Sixers have been cosmically intertwined. For the season's first two months, they were the darlings of statistical analysts. A series of blowout wins over lesser foes helped both teams compile gaudy point differentials, but their inability to win close games kept their records from matching up with their numbers.
Since then, both Portland and Philadelphia have crashed. The Blazers quit on Nate McMillan after the All-Star break, which resulted in an overhaul at the trade deadline that sent two starters elsewhere and saw McMillan replaced on an interim basis by Kaleb Canales. Portland has played .500 basketball under Canales, but has quietly slipped out of the playoff race nonetheless. Philadelphia's fall has been somewhat gentler. The Sixers built enough of a cushion in the Eastern Conference that they could survive nine losses in 13 games. Thanks to wins over lottery-bound New Jersey and Toronto, Philadelphia appears to have arrested the slide enough to likely make the playoffs. However, an Atlantic Division crown that once looked like a formality proved nothing more than a mirage.
The 76ers and Blazers are two of the most extreme examples of a league-wide trend toward massive swings during this lockout-shortened season. Splitting the season at the All-Star break, the correlation between schedule-adjusted point differential in the first and second halves is .687. That's down substantially from recent seasons:
Surely, much of the explanation for the extremes we've seen this season is due to nothing more sinister than sample size. Cutting the schedule by a fifth reduces the number of games before the break, and we're not yet through the second half. There were similar changes during the last lockout-shortened season. Using splits from DougStats.com, we can calculate the correlation between point differential in both halves of the 1998-99 season, which comes out at .684. That's higher than this year's correlation, unadjusted for schedule (.618), but much lower than the typical unadjusted season (about .8).
To provide some context for what this means, typically the variation in performance before the All-Star break explains more than 70 percent of the same variation after the break. This year, it explains less than 50 percent.
Another way to consider how much teams have varied over the course of the season is to look at the most extreme examples. Surprisingly, the Sixers don't qualify as one of the five largest declines. While their point differential is down from +6.5 points per game before the break to +1.2 since, more than half of that difference is attributable to a far more challenging schedule. In terms of adjusted performance, Philadelphia's drop from +5.1 points better than average per game to +2.8 ranks just eighth in the league.
Here are the top five in either direction, including Portland:
Team Pre Post Diff Team Pre Post Diff
New York -0.9 +6.4 +7.3 Portland +5.0 -5.5 -10.5
Phoenix -2.2 +4.4 +6.6 Dallas +4.3 -2.1 -6.4
Washington -9.0 -3.4 +5.7 Minnesota +1.0 -4.0 -5.0
Boston -0.7 +4.5 +5.2 Miami +8.3 +4.5 -3.9
Milwaukee -1.7 +3.1 +4.8 Cleveland -4.8 -8.0 -3.2
The change in the Blazers' performance is the largest in the five seasons for which I have adjusted game-by-game ratings. In fact, prior to this season, just three teams (the 2009-10 Lakers, 2008-09 Celtics and 2007-08 Clippers) declined by more than five points per game. Both Portland and Dallas have suffered larger drop-offs than any of their predecessors during this span.
There was a similar extreme difference in 1998-99, as the Chicago Bulls went from bad during the first half of the schedule to historically terrible over the last 25 games, during which they were outscored by an unthinkable 14.4 points per game. Their drop between the halves was 10.0 points, nearly as large as this year's Blazers.
Large improvements have been slightly more common in non-lockout seasons because of the opportunity to add talent at the trade deadline. The biggest positive change in this period was last year's Denver Nuggets, who improved by 7.6 points per game after the deadline, slightly more than this year's New York Knicks. (Oddly, the two teams are tied by Carmelo Anthony.) Three teams (the Indiana Pacers, New Jersey Nets and Phoenix Suns) improved by at least six points per game during the second half of 2009-10.
The value of splitting the season in two is more difficult to determine from the past. Note that the 2009-10 Lakers turned the switch on for the playoffs and won the second of their back-to-back championships, beating a Boston team that also slumped after the break. Of the teams that improved after the All-Star break, only the 2010 Suns really made much noise in the postseason.
Going back to the last lockout doesn't offer much more clarity. The eventual champion San Antonio Spurs excelled during the second half, going 21-4 with far and away the league's best point differential. However, the Knicks' unprecedented run to the NBA Finals from the No. 8 seed contrasted with their poor play in the second half. New York actually declined by 3.2 points per game in the second half of the season.
Teams like Dallas, Miami and Philadelphia that have slumped in the second half, but not badly enough to knock them out of the playoffs, will hope to repeat the Knicks' unpredictable surge this time around.
Kevin Pelton is an author of Basketball Prospectus.
You can contact Kevin by clicking here or click here to see Kevin's other articles.