Not As Efficient As You'd Think
By crakpot, 8th Aug 2010 | Follow this author
| RSS Feed | Short URL http://nut.bz/1ijh65t5/
Posted in WikinutNewsSport
What's up with the NBA's Efficiency Rating? Learn how to find each player's score, and learn a simpler, infinitely more valuable statistic to find a player's contribution to his team.
Not As Efficient As You'd Think
Yesterday, I tackled the NFL's quarterback rating (you can read the article here), and replaced it with a simpler, and more effective, measure: Win Score. Today, I take on a similar measure in the NBA: Efficiency Rating.
For those unfamiliar with this stat, it is fairly easy to calculate:
NBA Efficiency = ((Points + Rebounds + Assists + Steals + Blocks) - ((Field Goals Att. - Field Goals Made) + (Free Throws Att. - Free Throws Made) + Turnovers))
Looks simple, right? Basically, you add up all of the stats on a player's scoresheet, and that number is the player's efficiency score.
Unfortunately, there are many innate flaws in this rating system. Because of the way the formula is calculated, a player with averages far below the NBA average can increase his score as long as he keeps shooting. To break even, a player must have a field goal percentage of .333, a 3-point percentage of .250, and a free throw percentage of .500. A player with these numbers would certainly not be seen as adding value to his team, yet, according to this formula, he would do just that. Also, why is a field goal attempt worth just as much as a free throw? Are blocks really worth as much as points or rebounds?
We generally think of the NBA as a smart organization. However, the fact that they use this outdated metric to value game performance lends us to believe otherwise. Fortunately, there is another metric available, one just as simple to compute, yet infinitely more valuable to fans and coaches alike.
This measure, developed by David J. Berri, Martin B. Schmidt, and Stacey L. Brook in their groundbreaking book The Wages of Wins, is called Win Score, and it is computed as follows:
Win Score = Points + Rebounds + Steals + 0.5*Assists + 0.5*Blocks - Field Goals Attempted - Turnovers - 0.5*Free Throw Attempts - 0.5*Personal Fouls
Under this system, the minimum averages that a player could have and still gain points are .500/.333/.500 (FG%,3P%,FT%). Per player, the average Win Score per Game was 5.33. However, the ultimate test of efficiency is WS48, or Win Score per 48 minutes. This can be computed via a very simple formula:
WS48 = 48*Win Score / Minutes Played
Among players who played at least 1000 minuted last year, Marcus Camby, the power forward for the Portland Trail Blazers, leads the NBA in WS48. However, this raw stat is heavily biased towards forwards and centers, who make up the top 10 spots on the list.
Fortunately, there exists a simple solution to this problem. We can take the average Win Score at each position, and then weigh the positions accordingly. Doing this, we learn that the modifiers for each positon are:
Center 0.71
Forward 0.86
Guard 1.38
After adjusting our stat to meet this new data, we learn that the most efficient player last year was... (drumroll please) Jason Kidd! (gasp!) Just like we did with QB Score, we can assign grades (A+, B-, etc) based on WS48. Using this data, we can grade the most hyped team this offseason, the Miami Heat, on their starters' performance last season:
Starters Position wWS48 Grade
Mario Chalmers G 4.16 F
Dwyane Wade G 15.35 A+
LeBron James F 14.84 A+
Chris Bosh F 14.54 A
Joel Anthony C 5.63 F
Bench
Carlos Arroyo G 8.17 C-
Eddie House G 5.42 F
Mike Miller G 15.32 A+
Udonis Haslem F 10.94 B+
Juwan Howard F 5.37 F
Zydrunas Ilgauskus C 6.08 F
Notes: Mike Miller is FAR too good to be a bench player; if Dwyane played point guard, then Miller could take the other guard spot, and the lineup would have four grade-A starters... Mike Miller is the strongest challenger to Tim Duncan for the top white player in the Association.... If Udonis Haslem (6'8", 235 lbs.) played the center position, then the line-up would look like this (with Mike Miller): A+, A+, A+, A, B+. Just food for thought.
Full wWS48 data is available here.
Comments
9th Aug 2010 (#)
Interesting and does make sense with two exceptions. The bias towards forwards and centers and also it doesn't take defensive stats into consideration. A strond defense is a very valuable asset.
Reply to this comment
9th Aug 2010 (#)
I thought about that. Unfortunately, many defensive stats are not available, and matchup data (who guards whom, etc) is not kept by the NBA. Defensive measures that are used in the formula are defensive rebounds, blocks, and steals.
Reply to this comment