A note about PER (Player Efficiency Rating)

As I'm going to be using the PER statistic to describe player performance, I thought a quick note about it would be of interest to some of you. PER was developed by John Hollinger, the preeminent basketball statistical analyst in the game. Hollinger wanted a way to quantify player achievement for diverse skill sets. That is he wanted to put a numerical value on each players contribution to the game, independent of minutes played. This is a fairly radical concept, as usually we compare shooting guards to shooting guards (Kobe scored 32 last night, Pierce only scored 24), and centers to centers. By being able to compare the impact of one players ability to make assists to another players ability to grab rebounds, lineups, trades and free agency can all be evaluated more fairly.

The PER is a complex formula based on a variety of statistical factors. As Hollinger explains in his book, Pro Basketball Prospectus (published yearly before Hollinger went to work at ESPN),
"the PER rates everything except postion defense." While a variety of factors play into the formula, a great deal of it is based upon using the value of a possession, which is simply based on the average points scored for each possession in the NBA. This is approximately 1.02 points/possession for a given year though it does flucuate. He then scales statistics accordingly based on whether they add possessions (i.e. rebounds, steals, blocks) or give the other team possessions (missed shots, turnovers, fouls). If a player does something to give his team a possession, and a possession is on average worth 1.02 points in any given game, then it makes sense to quantify this as being worth 1.02 points.

The remainder of the PER calculation is based on the ability of each player to score. The value of a 3 pointer is obviously worth 3 points. Hollinger goes in to detail to find exact values for 2-point field goals and free throws (which are more complex than the 3-pointer data), and even quantifies the value of an assist in regards to points scored. The amount each player scores is then scaled by a Pace Factor, which takes into account the fact that teams that play faster will score more points and teams that play slower will score less points. Obviously a team like Phoenix or Denver will have a higher pace factor, while a more methodical halfcourt team like the Bulls or the Heat will have a lower pace factor.

Hollinger then takes the final data, and sets the league average to be 15, and adjust the number accordingly. So an average player will have a PER of 15. Hollinger came up with this table to give a feeling for the relative value of a specific PER

*A Year For the Ages: 35.0
*Runaway MVP Candidate: 30.0
*Strong MVP Candidate: 27.5
*Weak MVP Candidate: 25.0
*Bona fide All-Star: 22.5
*Borderline All-Star: 20.0
*Solid 2nd option: 18.0
*3rd Banana: 16.5
*Pretty good player: 15.0
*In the rotation: 13.0
*Scrounging for minutes: 11.0
*Definitely renting: 9.0
*On next plane to Yakima: 5.0

Here are Hollinger's calculated 2006-2007 season PER's for the Sonics on the roster for this year (note as there is no data for Rookies, they are unranked).

Nick Collison: 14.26

Mickael Gelabale: 9.85

Johan Petro: 11.63

Luke Ridnour: 13.74

Mouhamed Sene: 8.47

Robert Swift: 12.75 (from 2005-2006 season due to injury last year)

Wally Szcerbiak: 14.90

Kurt Thomas: 11.37

Earl Watson: 13.58
Delonte West: 13.50

Chris Wilcox: 16.63

Damien Wilkins: 13.13

Well, there it is. The brutal truth. The Sonics have 1 (ONE) player on their roster who was an above average NBA player last year (aside from positional defense). It's not surprising to see that Chris Wilcox rates as the best returning Sonic, he's the only player who was anything approaching statistically impressive and he played alot of minutes. But wait, PER in minute independent, meaning the Sonics aren't hiding anyone on their bench who is likely to break out with increased playing time. The low PERs on the players with NBA experience shouldn't be surprising, as the Sonics lost their 2 best players from last year, Rashard Lewis due to free agency (though the Sonics squeezed out a useful trade exemption which they used in the Kurt Thomas trade) and through the Ray Allen trade. Obviously, Kevin Durant will likely end up being a better than average NBA player this year (and much better in the future), and hopefully we'll see the same out of Jeff Green.

It also allows us to begin to evaluate the Ray Allen trade. Ray Allen had a PER of 21.70 last year, just shy of being a borderline all-star. He's played at the same level for 5 years now with an average PER of 21.62, remarkably consistent. He's also 32 years old, and though the delcine hasn't started yet, its fair to expect it to start soon. The Sonics also sent the 5th pick in the 2nd round (#35 overall) to the Celtics. In return, the Sonics recieved Wally Szczerbiak, Delonte West, and the 5th overall pick in the 2007 NBA draft which the Sonics turned into Jeff Green. Szczerbiak had a PER of 14.90, almost exactly league average, and West had a PER of 13.50, a rotation-level player. Szczerbiak is 30, and has already started to decline, so he shouldn't be expected to produce at a higher level over the remainder of his contract, West however, is only 24, and its not unreasonable to expect him to develop into an average-slightly above average NBA player.

Green of course is the wild card, which determines whether or not the trade is in favor of the Celtics or the Sonics. There are two ways of looking at his value. The more complex way involves attempting to project his ability level based on his college stats. Hollinger did exactly that (subscription required for link), and determined that Green was the 14th best player in the draft, which would mean the Sonics dramatically reached for Green. According to Hollinger, better college players which the Sonics should have taken include Thaddeus Young (taken in the middle of the first round), Brandan Wright (taken 8th overall), and Nick Fazekas (taken in the 2nd round). Fazekas is the perfect example of a player that splits stat-guys vs. scouts. Hollinger had him rated as the 7th best college player in the draft. Due to his lack of athleticism, he was taken in 4th overall in the second round.

The other method to determine Green's value is to look back at the PER of players drafted 4th overall the past few years. I'm only using players drafted the last 5 years (this is a completely arbitrary cut-off, data is just easier to get for the last 5 years.

2006: Shelden Williams

2006-2007 PER: 12.39

Rookie year PER: 12.39 (last year was his rookie year)

2005: Raymond Felton

2006-2007 PER: 13.55
Rookie year PER: 14.30

2004: Devin Harris

2006-2007 PER: 15.96
Rookie year PER: 14.76

2003: Dwyane Wade

2006-2007 PER: 17.61
Rookie year PER: 29.04

2002: Nickoloz Tskitishivili

2006-2007 PER: Out of NBA
Rookie year PER: 4.90

It's tough to get an average PER out of this mess, so let's just look at player types. 1 player is an absolute superstar, 1 is an above average point guard on a championship contender, two others are rotation players, and one is out of the NBA completely. Based off these 5 guys (an extremely small sample size) you get a 40% chance of getting an above average player, a 40% chance of getting a rotation player, and a 20% chance of getting a complete bust.

As always there are almost infinite amounts of other information that can go along with this. Green is extremely unlikely to be a bust due to his college history ('Skita' was almost a complete unknown). He's also extremely unlikely to become a Dwyane Wade-level superstar. Based off of the past 5th overall draft picks would seem to be a PER of 18-20.

If Green develops to that level, and Delonte West turns out to be an average NBA player, the Sonics will have traded old for young, and should get several years of productivity albeit at a slightly lower level than Ray Allen might produce for the next two years. This trade made immense sense for both teams.