Thursday, February 21, 2008

Adjusted Defensive Efficiency

The Hawks have looked really bad defensively lately. To the eye, the games @Phoenix, @LA Clippers, @Houston, @LA Lakers, @Sacramento have been defensive debacles. All are road games and all are against Western Conference teams. Neither condition is conducive to maintaining a high level of defensive performance. Per Knickerblogger's stats page, the 5 most efficient offenses in the NBA all play in the Western Conference, as do 6 of the top 8 and 10 of the top 15.

The Hawks were 7th in the league in defensive efficiency at the All-Star Break. After the losses to the Lakers and Kings, they're 13th. My question: Are the Hawks playing worse defense or are they playing better offenses? The answer is almost certainly both, but I've spent the better part of the day estimating a measure of this which does not seem to be fundamentally flawed. (Fundamental flaws should be pointed out in the comments.)

To get the offensive efficiency (points per 100 possessions) for each game I estimated possessions (fga + (.4*fta) - (1.07*or) + to) for both teams and halved the sum. Doing this for the whole season, my season averages for the Hawks (scoring 103 pts/100 poss and allowing 105.5 pts/100 poss) came closer to matching Aaron Barzilai's numbers at basketballvalue.com (scoring 103.2 and allowing 105.9) than Knickerblogger's (scoring 105 and allowing 107.7) or basketball-reference.com's (scoring 103.7 and allowing 106.4).

Hoping to come as close to possible in comparing like-to-like, I used Aaron's season averages for each opponent to determine how much better or worse than their season average they performed offensively against the Hawks. I used each opponent's season average including their game(s) against the Hawks. I don't know if this is a serious mistake or not, so, again: Fundamental flaws should be pointed out in the comments.

For example, on the year Sacramento scores 106.3 points per 100 possessions. Last night they scored 124.8 points per 100 possessions. That's 117.4% of Sacramento's average offensive efficiency. That's bad. That's Atlanta's worst defensive performance of the year by this measure. (The second worst: 115.9% at Houston on February 9th. The best: 80.8% against Memphis on December 8th.)

During the current six game losing streak, opponents have been more efficient than their season average.

OppOff Eff (Season)Off Eff (Game)Adj Eff
CLE105.3108.7103.2%
@HOU104.9121.6115.9%
DET110.3110.8100.5%
@CHAR102.7106.6103.8%
@LAL111.1119.8107.8%
@SAC106.3124.8117.4%

Everybody's having a good offensive night against Atlanta right now (though Detroit's offense was mostly just its usual very good self). Following the trend further back the schedule, opponents have been more efficient than their season average in 10 of Atlanta's last 12 games. The Hawks have lost 9 of those games. The New Jersey game is the only game in that stretch where Atlanta beat an opponent that bettered their season average offensive efficiency.

On the year, 36 of Atlanta's 51 games the result has matched Atlanta's adjusted defensive efficiency. (70.6% of the time Atlanta wins when they hold their opponent below their season average or loses when they don't.) Atlanta is 16-8 when their adjusted defensive efficiency is below 100. They are 5-23 when it is above 100. The Hawks have won just 1 of the last 15 games in which they've failed to hold an opponent below their season average offensive efficiency.

This is essentially (I believe) what Ken Pomeroy does in adjusting for competition between NCAA teams. I'm sure he does it more thoroughly and exactly. Obviously there's not the gulf of talent separating the NBA's best team from its 30th as there is separating the best Division 1 team from the 341st best, but has Atlanta's defensive efficiency rating benefited from playing the bulk of their games against Eastern Conference teams?

Summing the adjusted efficiency numbers for each of the 51 games (unfinished/re-opened business against Miami excluded) and dividing by 51 suggests that...the Hawks are slightly above average defensively overall. Just as their 13th ranked defense would suggest. It's not an illusion created by the schedule. Opponents have operated at 99.6% of their average offensive efficiency against the Hawks this season. Recent opponents are faring far better than that.

I did the work, though, and thought I'd share.

Ballhype: hype it up!

No comments: