Add it all up and you get ... the Player Efficiency Rating
Editor's note: To introduce our new feature, "Hollinger Player Profiles" for every NBA player, we asked John Hollinger to explain his basic measure of player performance: the Player Efficiency Rating. Click on a team or player name to see the profile for every player.
Productivity -- that's the name of the game in the NBA. Make more positive plays and fewer negative ones than your opponent, and there's a good chance your team will win the game.
Thus, a good way to measure players against each other is to compare how many positive and negative plays they make. To do so, I developed a system called the Player Efficiency Rating, or PER.
PER takes into account all of a player's statistical contributions -- points, rebounds, assists, steals, blocks, missed shots, turnovers and fouls -- and rolls them up into a score for every player in the league.
PER relies on a complex formula to rate a player's contribution in each of these categories, awarding or subtracting points depending on how much the player added or subtracted to his team. But it doesn't stop there. It adjusts every player's rating for the team's pace, an important consideration when comparing players from, say, Indiana and Phoenix.
Perhaps more important, it is a per-minute rating, not the per-game averages we're used to seeing. This distinction is important because we can then compare players whose playing times may be very different and answer questions such as "Should Player X play more than Player Y?" or "Could Player B be helpful to Team A?"
To simplify things, I set the league average in PER to 15.00 for each season. This makes comparisons between seasons very easy, which affords other opportunities. For instance, we can evaluate a player against his previous seasons, compare players from different eras, or produce objective answers to questions such as, "Are there more great power forwards now than at any time in history?"
Furthermore, using 15.00 is a helpful benchmark because that's about what we expect a run-of-the-mill starter to score in an NBA game. Thus, we can use barometers already hard-wired into our heads -- above 20 is excellent, above 25 is superstardom, and below 10 is brutal.
Finally, there is one partial blind spot: defense. PER analyzes players based on blocks, steals and fouls, but that's a small part of overall defense. As a result, defensive specialists who get few blocks or steals, such as Bruce Bowen, end up with a lower rating than we might expect. Similarly, bad defensive players who nonetheless pile up steals or blocks, such as Jason Williams or Raef LaFrentz, rate higher than they ought to.
So, let's get to the fun part. Just who were the most productive players last season? The chart below shows the NBA's top 10 in PER last season. I also inserted each player's per-40-minute stats to show how each ended up on the list:
|PER: Top 10 (minimum 500 minutes)|
|Kevin Garnett, Minnesota||23.3||14.2||6.0||28.35|
|Tim Duncan, San Antonio||24.4||13.3||3.3||27.13|
|Shaquille O'Neal, Miami||26.8||12.2||3.2||26.95|
|Amare Stoudemire, Phoenix||28.8||9.9||1.8||26.69|
|Dirk Nowitzki, Dallas||26.9||10.0||3.2||26.18|
|LeBron James, Cleveland||25.7||6.9||6.8||25.75|
|Andrei Kirilenko, Utah||19.0||7.6||3.9||24.45|
|Kobe Bryant, L.A. Lakers||27.1||5.8||5.9||23.28|
|Allen Iverson, Philadelphia||29.0||3.8||7.5||23.23|
|Yao Ming, Houston||23.9||10.9||1.0||23.22|
Garnett's team didn't make the playoffs, but he was the highest-rated player for a second straight season. Garnett ranked ahead of players like Duncan, Shaq and Stoudemire primarily because he stuffed the stat sheet in so many ways. While he averaged fewer points per minute than the next six players, Garnett was better at nearly everything else. He had one of the highest rebound rates in basketball, with only Duncan coming close among the top five in PER. Garnett had nearly twice as many assists per minute as the others and was well ahead in steals too. And compared to Duncan, his only rival in rebounding, Garnett shot better both from the floor and the free throw line.
Sum it all up and it's no surprise that Garnett rated No. 1; it's just unfortunate that the team around him fell apart, denying him a second straight MVP award. Speaking of MVPs, you might notice somebody isn't on the list. Phoenix's Steve Nash won the award but ranked only 18th in PER. While his assist rate and shooting percentages were phenomenal, only two players in the top 40 (Jason Kidd and Brad Miller) scored less frequently, and Nash didn't help his cause in the rebounding and defensive categories.
OK, we've looked at the best now how about the worst?
|PER: Bottom 10 (minimum 500 minutes)|
|Theron Smith, Charlotte||8.2||9.1||2.2||5.10|
|Mark Madsen, Minnesota||5.8||8.5||1.2||6.76|
|Rafael Araujo, Toronto||10.6||10.0||0.9||6.87|
|Junior Harrington, Indiana||11.9||4.6||4.5||7.63|
|Ryan Bowen, Houston||7.4||5.0||1.2||7.65|
|Calbert Cheaney, Golden State||10.5||5.2||2.7||7.76|
|George Lynch, New Orleans||7.0||7.5||3.8||7.83|
|Jarron Collins, Utah||8.9||6.8||2.5||7.92|
|Erick Strickland, Milwaukee||11.9||4.1||4.5||8.01|
|Walter McCarty, Boston-Phoenix||11.5||6.1||1.6||8.08|
Catch them while you can, folks, because you won't be seeing most of these guys in the NBA for much longer. One glaring exception is Madsen, the league's second-least productive player a year ago, who inexplicably received a five-year deal from the Wolves in the offseason.
Looking at last year's best and worst is informative, but we can also take things a step further. By comparing a given player to the most similar performers from recent NBA history, and then seeing how those similar players fared in subsequent years, we can project how a player's performance is likely to change in future seasons. Using just such a method, I have projected stats for every regular player, which you can find in the Insider section of the player stats. Based on those projections, we can determine the league leaders in PER for 2005-06.
Looking into the crystal ball, we shouldn't expect the top 10 to change much. Again, KG is on top:
|PER: Projected 2005-06 Top 10|
|Kevin Garnett, Minnesota||22.7||14.0||5.1||26.39|
|Tim Duncan, San Antonio||23.4||13.2||3.3||25.12|
|Shaquille O'Neal, Miami||22.1||11.9||2.7||24.24|
|LeBron James, Cleveland||26.7||6.2||7.1||24.09|
|Yao Ming, Houston||23.6||12.2||1.7||24.05|
|Tracy McGrady, Houston||27.3||6.1||5.7||24.03|
|Dirk Nowitzki, Dallas||23.9||9.6||2.9||23.52|
|Amare Stoudemire, Phoenix||25.9||10.2||1.9||23.51|
|Kobe Bryant, L.A. Lakers||25.7||5.7||5.7||22.46|
|Dwyane Wade, Miami||23.9||5.2||6.2||22.39|
Projections have their weaknesses -- for example, they don't know that Stoudemire just had microfracture surgery. But as a forecasting tool, projections can point us in unexpected directions. For instance, the two Rockets, McGrady and Yao, both project to be more productive than they were a year ago. If true, that would make Houston a much more serious threat in the West than many have anticipated.
But in the big picture, the triumvirate that has been ruling the league for nearly the entire post-Jordan era -- Garnett, Duncan and O'Neal -- again figures to lead the pack. That's hardly surprising -- they've accounted for four of the past six MVP awards and six of the past seven championships. Unless 2005-06 becomes Year 1 of the LeBron Era, look for the trio to make it 5 of 7 on the MVPs and 7 of 8 on the championships.