-- Based on performance during the regular season, the 2015 NBA Finals are a mismatch. The Golden State Warriors (67-15) won 14 more games than the Cleveland Cavaliers (53-29), the largest differential since the 67-win Boston Celtics took on the 51-win Houston Rockets in 1986.
So far, however, the postseason has told a different story. It's the Cavaliers, not the favored Warriors, who enter the Finals the hotter team. Cleveland went 10-2 during the Eastern Conference playoffs, tying the 2003 New Jersey Nets and 2013 San Antonio Spurs for the fewest number of games needed to reach the Finals since the opening round went to best-of-seven. And even accounting for the weaker competition in the East, the Cavaliers have still been the better team in the playoffs.
Evaluating playoff performance
To adjust for widely varying quality of opposition in the playoffs, I like to simply compare teams' point differentials to the average point differentials their opponents posted during the regular season. Weighted by games played in each series, Golden State's opponents posted a plus-2.6 point differential during the regular season.
The Warriors have actually beat those teams by 8.0 points per game, meaning they've been 10.6 points better per game than an average team -- slightly better than their plus-10.1 point differential during the regular season.
That's good, but Cleveland has been better. Believe it or not, the Cavaliers have actually faced tougher opposition en route to the Finals. Their average opponent had a plus-2.9 differential in the regular season, and Cleveland has beaten those teams by 8.8 points per game, making the Cavaliers 11.7 points per game better than an average team.
Now, an obvious counterpoint is that Cleveland didn't have to beat the same Atlanta team that had a plus-5.4 point differential during the regular season because of injuries to Kyle Korver and Thabo Sefolosha. To adjust for that, I also used the multiyear version of ESPN's real plus-minus (RPM), weighted by minutes played, to estimate the actual talent of the opponents both teams played on their way to the Finals in terms of net points per 100 possessions.
By this method, Golden State did have a tougher path to the Finals, in large part because the Hawks were so badly hampered by injuries. (Atlanta actually rates by weighted multiyear RPM as the weakest opponent either team faced along the way.)
And from this perspective, the Cavaliers were at their best in sweeping the Boston Celtics in the conference quarterfinals with Kevin Love playing three of the four games. Still, Cleveland's playoff run rates better overall.
Improved Cavaliers defense
We can perform a similar analysis looking strictly at offense and defense. Using regular-season performance, the Cleveland offense has scored 7.1 more points per 100 possessions than we'd expect an average team to score against its playoff opponents. (The weighted RPM model has a similar estimate.) That's impressive, given that the Cavaliers have been without Love and Kyrie Irving has been in and out of the lineup due to injury. Still, it's slightly worse than the Cavaliers managed after their trades and LeBron James' return from a midseason absence. They were 7.5 points per 100 possessions better than league average from Jan. 13 onward, per NBA.com/Stats.
The larger Cleveland improvement has come at the defensive end. The Cavaliers were a below-average defensive team during the regular season, ranking 20th in defensive rating, and were only slightly better than average (plus-0.3 points per 100 possessions) after Jan. 13.
In the postseason, Cleveland has the league's third-lowest defensive rating, and even adjusting for the poor offenses they've played against, the Cavaliers have allowed 5.9 fewer points per 100 possessions than we'd expect based on their opponents during the regular season. (Again, the RPM model says something similar.)
Cleveland's total improvement of 6.9 points per 100 possessions in defensive rating from the regular season to the playoffs is one of the best by any team that reached the conference finals since the playoffs went to 16 teams in 1984.
The Cavaliers have done it with a different defensive style than the regular season. They're forcing fewer turnovers -- the second-lowest rate of any team in the playoffs, in fact -- but also fouling less, grabbing more defensive rebounds and most importantly holding opponents to the lowest effective field goal percentage of anyone in the postseason.
It's tempting to attribute the difference to Tristan Thompson replacing the injured Love in the lineup, given Love's poor defensive reputation. But again, the weighted RPM measure says the Boston series was Cleveland's best on defense. The Celtics scored 8.4 fewer points per 100 possessions than expected based on their lineups. Against the Bulls and Hawks, the Cavaliers' defense was 5.3 points per 100 possessions better than expected.
Instead, the bigger factor has been opponents' inability to make 3s against Cleveland. Teams have shot 28.1 percent from 3-point range against the Cavaliers, by far the lowest mark of any playoff team (Golden State is second at 31.0 percent). There's ample statistical evidence that teams have only limited control over how well their opponents shoot from beyond the arc, particularly in small samples.
A counterargument is that teams with poor opponent 3-point shooting are leaving the right players open. Seth Partnow of Nylon Calculus has debunked that by showing that players have shot worse on wide-open 3-point attempts (no defender within six feet, as measured by SportVU camera tracking) during the playoffs against Cleveland than they did overall during the regular season.
Nearly a quarter of the Cavaliers' defensive improvement can be attributed to opponents shooting below average against them from 3-point range. So even without it, Cleveland would be much better, but some of the improvement might not prove sustainable in the Finals.
Value of "peaking at the right time"
So how important is the Cavaliers' playoff superiority? Last season, when predicting the Miami Heat-San Antonio Spurs matchup, I took a look at what helped determine the outcome of past Finals. Previous playoff performance adjusted for opponent was one of the key factors. Lower-seeded teams like Cleveland that have outperformed the higher-seeded team in the playoffs have gone 6-8 in the Finals since 1984 (.429), as compared to a 2-15 record (.118) for underdogs who were weaker in the first three rounds of the playoffs.
Of course, the lower-seeded team sometimes had a much bigger playoff advantage than this season's Cavaliers, and underdogs with an edge between 0 and 2 points per game like Cleveland (plus-1.1 compared to Golden State) have gone 2-4 (.333).
Regular-season performance is still the dominant factor in predicting the outcome of the Finals, and even with the Cavaliers' playoff edge the regression I came up with last year gives the Warriors a 98 percent chance of winning the series. That overstates Golden State's edge because Cleveland has been so much better after its midseason trades.
If we instead use the 63-win pace at which the Cavaliers played after Jan. 13, the series looks much more competitive, with the regression giving the Warriors just a 56 percent chance of winning -- somewhat less than the gambling odds would indicate. If we take out the house edge, the lines at the Westgate Las Vegas SuperBook (5-12 for Golden State, 2-1 for Cleveland) imply the Warriors are expected to win about 68 percent of the time.
There certainly is something to how the Cavaliers have played during the playoffs, then. It just might not be enough to overcome a Golden State team that is far and away the best team Cleveland has played this postseason.