Stat Central: Kevin Durant’s In-Season Decline

Kevin Durant had by far the best season of his career. He won his first MVP trophy, posted the highest PER mark of his career (29.8) and accumulated 19.2 win shares–19.2 win shares is a monster number and the seventh-best mark in the since the NBA has had a 3-point line.

The only players to have a statistically better season since the 1979-80 season are: Michael Jordan (four times), LeBron James (twice) and David Robinson (once).

He carried a young supporting cast when Russell Westbrook was injured to a 22-10 record, a winning percentage that would have placed the Thunder as the No.1 seed in the East. During a three-month period from January to March, he averaged an insane 34.7 points per game.

Durant did all of this while simultaneously improving his defense and becoming elite at that end.

Durant wasn’t just good, he was extraordinary.

Yet the playoffs and the last month of the regular season was a bit of a different story. Durant’s production dipped, and even though he still put up pretty good numbers, he just wasn’t the same.

First 72 Games: 32.2 Pts, 51.4 FG%, 42.4 3P%, 7.6 Reb, 5.6 Ast, 2.1 Stl + Blk, 25.2 GmSc
Last 29 Games (including playoffs): 30.0 Pts, 45.9  FG%, 31.7 3p%, 8.1 Reb, 4.0 Ast, 2.0 Stl + Blk, 20.8 GmSc

Now that may not seem like much, and you might think it could be explained by just a few bad games and tougher competition in the playoffs. But players only tend to decline about 1.5 percent from the regular season to the playoffs statistically, most of that comes from shooting percentages going down, and it’s often offset by other things such as higher rebounding rates.

Here are all Kevin Durant’s games as a linear regression chart, illustrated by Game Score (basically the box score in a number) per 48 minutes.

That’s some serious decline towards the end of the season, and it’s either a 30-game slump (probably a part of it) or something else going wrong (likely a larger factor).

Durant played 3,122 minutes during the regular season, which was the most in the league by a difference of a hundred minutes and 12 to 17 percent more than the likes of Kevin Love, Goran Dragic,  James Harden and Lance Stephenson. Add to that more than 800 playoff minutes and you’ll start to get an idea of how much Durant has to do for this Thunder team.

Durant is only 25, and sometimes we like to think of these guys as invincible. But it’s possible that Durant’s workload was just too much and he started wearing down towards the end of the season.

Just going by the eye-test, when Tony Allen was pestering him during their first round series against the Grizzlies, it sure looked like Durant was worn down.

In the last 10 years, there have been only four other comparable seasons by minutes and usage rate: Kobe in 2005-06 and 2006-07, LeBron in 2005-06 and Allen Iverson in 2004-05. Compared to them Durant scored more points, had more win shares, shot a higher percentage and had a much higher sum of steals, blocks, rebounds and assists.

Durant had the most taxing season (at least statistically) since Michael Jordan in 1989-90. Insane!

Scott Brooks is often faulted for not playing his best players and lineups (*cough* Kendrick Perkins *cough*), but it’s possible that the most horrendous thing he is doing is running the face of his franchise to the ground. Based on the evidence it’s too early to say that, but Durant is poised to play in this summer’s World Cup, meaning he won’t have much of a summer holiday.

It’ll be interesting to see if Durant is forced to carry a historically large load again and whether or not these trend lines repeat again next year.

It’s certainly something to keep en eye on.

*Chart Details: Correlation coefficient (r): -0.16470772665106. Regression line equation: y=32.944858020075-0.056194596046107x. Standard deviation (σ): 9.9936817539883. Mean y (ȳ): 30.075. Median: 29.6. Lowest value: 6.1. Highest value: 52.9. Range: 46.8. Interquartile range: 12.625. First quartile: 23.375. Third quartile: 36. Variance (σ2): 99.873674999999. Quartile deviation: 6.3125. Mean absolute deviation (MAD): 7.8375