Rating a Player’s Efficiencyseries of blogs about advanced basketball stats. Thus far we have discussed John Hollinger's Player Efficiency Rating (PER) stat and the plus/minus stat.
This week I'll talk about offensive and defensive rating (ORTG and DRTG). These measures were created by Dean Oliver in his groundbreaking book Basketball on Paper about a decade ago as a way to measure offensive and defensive efficiencies for either an individual or a team. For more detail on the numbers, hows, and whys of Oliver’s methods you should check out the book directly, but here I will give my interpretations on the stat’s usage and usefulness.
What it is: For an individual player, the basic idea of the rating is to count how many points were produced (ORTG) by the player or given up by the team per possession while that player was on the court (DRTG). The simple formula for ORTG is points produced * 100/possessions, and for defense is points allowed * 100/possessions. Since ORTG is an individual score, points produced by the individual are calculated from shots made, free throws, assists, and offensive rebounds. The DRTG is simpler, as it simply adds up how many points were scored while a player is on the court and normalizes for 100 possessions. The higher the ORTG the better, whereas the lower the DRTG the better.
Strengths: The ratings are simple and easy to understand, as there is very little hard math required to come up with them. The ratings are also intuitive, as efficient offense = high ORTG score and preventing opponents scoring = low DRTG scores just makes sense.
Measuring good defense directly is especially complicated, as picks/switches and fast breaks make it too difficult to pin one individuals points directly to the most responsible defender. Also, things like good rotations, strong shows on pick-and-rolls, and many other things that make up “good defense” just don’t show up in the boxscores. Thus, a measure like DRTG is a good catch-all for all of those intangibles.
Finally, because the ratings are normalized they help show the difference between efficiency and other tricks. For instance, Allen Iverson in his prime scored a lot but he took an awful lot of possessions (lots of shots and turnovers) to get those points. Meanwhile, Chris Paul has never approached leading the league in scoring but he shoots high percentages, dishes tons of assists and loses very few turnovers. Thus, we would expect Paul’s career ORTG (120) to be a lot higher than Iverson’s (105), which it is. This tells us that Paul produces 120 points per 100 possessions for his team, whereas Iverson only produces 105 points per 100 possessions for his team.
This normalization also clarifies the difference between a defense that makes it hard for opponents to score and one that allows low points/game due to slowing down the game. This is easiest illustrated using team defenses. For instance, in 2007-08 the Pistons gave up the fewest points in the NBA (90.1 ppg) with the Celtics second (90.3 ppg). But the Pistons played a much slower pace than the Celtics, so the Celtics actually made it much harder for their opponents to score on a possession-by-possession basis (Celtics’ DRTG = 98.9, Pistons’ DRTG = 102.9) and thus the Celtics really had the better defense overall.
Weaknesses: For offense, efficiency does not always denote best. Players that do not touch the ball often, and when they do touch it are often set up for easy shots (like an offensive garbageman or a spot-up shooter) may score a lot per 100 times they touch the ball but not have much offensive impact for their team (i.e. Carl Landry, currently fourth in NBA in ORTG or Andris Biedrins, league leader in ORTG for 2007-08).
On defense, because DRTG is a team measure it can be unduly influenced by teammates. If a great defender is surrounded by poor defenders his DRTG may be worse (i.e. higher) than others on other teams a comparable ability while a poor defender can get a boost by being in a good system. For instance, in 2007 Kevin Garnett’s DRTG was 101 while the average DRTG of the other four starters was over 110. In 2008 KG got adequate defensive teammates and led the league in DRTG (94), while those same four teammates from Minnesota (Ricky Davis, Mark Blount, Trenton Hassell, Mike James) all saw their DRTG rise (average DRTG = 113) once no longer next to Garnett. This was clearly a case where a great defender was being pulled down by weak defensive teammates, and correspondingly he was improving what should have been even worse scores for them.
Usage: You can’t simply say “player X has a higher ORTG or lower DRTG than player Y so he is better”. Instead, you say that player X is more efficient on offense than player Y, and then you look at other stats that measure things like usage, overall production (i.e. PER), and impact (i.e. +/-) and come up with a more complete picture of that player’s quality.
For instance, I mentioned above that Landry has a very high ORTG this season. But when you factor in that he is only playing 21 minutes per game and that his on-court/off-court +/- is negative 7.2 it becomes clearer that Landry is a role player and not an offensive juggernaut. Landry’s ORTG was a bit higher than Ray Allen’s, but investigating Allen’s minutes (37 per game) and on-court/off-court +/- (plus 15.8) shows that he in fact has a much larger impact than Landry. This matches our intuition for these two players. Once you start building up a library of advanced stats to use, it becomes easier to camouflage the weakness of any one measure by sanity-checking it against other measures as well.