The Rockets without Dwight Howard
Jan 19, 2015; Houston, TX, USA; Houston Rockets guard James Harden (13) talks with center Dwight Howard (12) during a Indiana Pacers timeout in the second half at Toyota Center. Rockets won 110 to 98. Mandatory Credit: Thomas B. Shea-USA TODAY Sports
James Harden put on a one man show against the Phoenix Suns last Tuesday with 40 points, 9 assists, 12 rebounds, 3 steals, and a block. He did all while his would-be partner-in-crime Dwight Howard looked on in an, as noted by friend of the site Charles Barkley, an equally impressive suit. Other than lacking some of the post moves of traditional bigs, Howard has seemed the perfect complement to Harden in a “Moreyball” system that keeps Harden slashing to the basket or behind the three-point line and Howard rolling out of pick-and-rolls like a man possessed.
This partnership has been short lived for the Rockets, with Howard sitting or playing limited minutes for around 40% of their season. Observing how this has affected the Rocket’s shot selection, and more narrowly Harden’s statline, has been made easier by Dwight’s injuries cutting up the Rocket’s season into four periods. The team can make adjustments that can set in over long periods when he misses multiple weeks at a time. Only their first game against the Warriors bucks this trend.
This is the Rockets before Dwight first extended absence. There’s been a lot written about the general shot selection of the Rockets, embodied by Harden and Howard. Next up are the Rockets in their most recent 12 games without Dwight. For the most expensive contract currently on the Rocket’s books, Howard’s absence does not jump out at you from the shot chart.
Even looking beyond the shots, missing Howard does not show up in the stat sheet. I grouped the Rockets’ season by whether or not Howard played and used two sample t-Tests for Equal Means to determine which stat categories changed. I did this for points scored, points scored against, FGA, FG%, 3PA, 3P%, FTA, FT%, ORB, TRB, assists, steals, blocks, turnovers, fouls, possessions (NC’s own), offensive efficiency, and defensive efficiency. The only categories that changed were free throws attempted (down from 26.31 with Howard to 22.15 without him, p = .065) and turnovers (down from 17.125 to 14.65, p = .035). Field goals attempted just misses, increasing from 82.5 to 85.6, p = .1085.
Looking at the team, Donatas Montiejunas has capably stepped into the center’s position. The eye-test certainly confirms this and Basketball-Reference says he’s spent 54% of his minutes playing at center and an increase in his minutes has a .28 correlation with Howard not playing. Josh Smith has also played 12% of his minutes nominally at the center position, more than at any point in his career. Further, for someone known as an unrepentant chucker, his usage percentage is almost exactly the same in Houston as it was in 2013-14 for Detroit. Per Seth Partnow’s Rim Protection statistics, Dwight Howard has allowed a 45.1% FG% at the rim, comparable to Montiejunas’ 46.9%, though better than Smith’s Detroit and Houston combined 50.5%.
With Howard, the Rockets have gone 21-11 and have been 15-6 without him, a higher win percentage that many have credited to the transcendent play of everyone’s least favorite MVP candidate James Harden. The narrative seems to imply that we should see a significant uptick in Harden stats, especially attempts, reflecting a general increase in usage and touches that would have been previously split with the other team’s superstar. Doing the same two sample t-Tests for Equal Means to compare Harden during the periods when Howard plays and does not we fine, somewhat surprisingly, no changes are statistically significant. The closest that comes to statistical significance are a slight increase in turnovers and a slight increase in steals, with p-values around 0.15.
Taken together, this form of analysis implies that the loss of Howard has not significantly impacted the box-score statistics of the Rockets outside of FTA and turnovers. The prevailing thought seems to have been that Harden has stepped up in a big way, but his numbers don’t support this. None of his numbers are statistically larger or smaller when Howard does not play. The conclusion I draw is that the combination of Donatas Montiejunas and Josh Smith banging down low as nominal centers and putting up serviceable rim protection numbers have made up for the deficit caused by Dwight’s absence.