Which MVP candidate is the most valuable in big games?

A statistical approach to determining the league's best player...

At the All-Star break, Russell Westbrook and James Harden are the top two candidates for the league’s Most Valuable Player award.

Since the departure of Kevin Durant, Westbrook has been carrying the Oklahoma City Thunder with a vengeance. He’s a human cannonball, exploding through defenses and sending rims to the emergency room. It’s no surprise that he’s leading the league in scoring. Most notably, Westbrook has achieved 27 triple-doubles this season. With 25 games remaining, he needs only 14 more to match Oscar Robertson’s record from 1961-62.

Meanwhile, James Harden has been putting in serious work in Houston. Harden has adapted to the role of point guard spectacularly, and is currently leading the league in assists at 11.3 per game. He has also pushed a defense-deficient Rockets team to third place in the Western Conference.

Both players have convincing cases for the MVP trophy this season, but only one man can win the award. I tried to add some clarity to this decision by figuring out which player was the most valuable in big games. In other words, whose play is least effected by the quality of the other team?

To measure this, I made an estimation of each player’s efficiency rating (PER) for every game this season, using a method provided by Bleacher Report. PER is not a perfect statistic, but it takes into account almost everything a player does during the game, weighting offensive performance slightly more than defensive performance. Then, I checked for a correlation between PER and the opponent’s win percentage. I expected that both players would have a negative relationship – it’s to be expected that the better the opponent, the lower the PER will be. Therefore, it is whoever has the PER least impacted by opponent win percentage that will be the most reliable in big games.

Running the numbers for Westbrook, I found that his PER does have a statistically significant negative relationship with opponent win percentage. This indicates that the better the team, the more inefficient Westbrook is likely to play.

(r = -.3, p = .02)

With Harden, I was expecting similar results. Harden’s PER, however, had no significant statistical relationship with opponent win percentage. The graphs look pretty similar because there is a decent amount of variability among individual games for both players, but on a season-long basis, it’s clear that Harden’s PER has a less sharp decline.

(r = -.17, p = .2)

This can only mean one thing: Long live the beard. Harden’s efficiency is less likely to be affected by the quality of team he plays, meaning that, on average, he has been a more valuable asset against playoff-caliber teams.

On another note, I found that Westbrook’s points per game did not have a significant relationship with quality of opponent. Despite a decline in his PER, Westbrook’s scoring tendencies stay relatively the same as opponent win percentage increases. This means that even if Westbrook plays inefficiently – which he has been known to do occasionally  – it is unlikely to affect his ability to score.

(r= -.07, p = .58)

Ultimately, there is still a lot of time left in the season, and both players have the opportunity to further prove themselves worthy of the league’s highest honor.