Wednesday, March 25, 2009
PER Diem: March 25, 2009
By John Hollinger
I get e-mails all the time about Player Efficiency Rating (PER), and while some of them are childish -- all-caps rants in which even the four-letter words are misspelled -- a lot of them are intelligent.
One of the most common themes among the latter group is that PER doesn't account for how many minutes a player has played; thus, while it's useful in a vacuum, it doesn't necessarily provide enough information for things like year-end awards and All-Star teams, in which how much a player does something is as important a factor as how well. This is particularly salient in this year's Rookie of the Year race, for example, in which the top rookie by PER, Marreese Speights of Philadelphia, averages only 15.3 minutes per game.
To help address that, it's time for a new acronym: VA. It stands for "Value Added," and the idea behind it is to measure not only quality but also quantity. Those of you familiar with VORP (Value over Replacement Player) in baseball will instantly recognize it, as it is the same type of measurement.
OK, that's three paragraphs and three acronyms, so let's try to get back to English.
The idea behind Value Added is to take the difference between a given player's performance and that of a "replacement level" talent -- the type of guy who might be sitting at the end of a team's bench, or perhaps in Sioux Falls -- and multiply that difference by the number of minutes that player played. The result shows, theoretically, how many number of points the player added to his team's bottom line on the season.
VA is very useful for award voting in particular, because it allows us to compare players with disparate production and minutes -- say, one player who was brilliant in 60 games and another who was merely good but played all 82 -- and figure out which performance was more productive.
Now for the nitty-gritty.
"Replacement level" is a common term in sports analytics for the level of talent readily available to be picked up off the scrap heap. In NBA terms, it's either a minimum contract guy at the end of the bench or some other team's bench, a veteran off the free-agent scrap heap or a call-up from the D-League.
Assigning a value to "replacement level" is perhaps more difficult in basketball than it is in baseball, as there is relatively little roster turnover during the course of a season.
Nonetheless, analyzing several years of data and looking closely at players who played less than 500 minutes in a season reveals a pattern. The average PER for a "replacement level" player in the NBA is about 11.0, and it varies by position. Power forward is the most replaceable position, with an average replacement level PER of 11.5; the two wing positions are the least, at 10.5 each; point guards, at 11.0, and centers, at 10.6, fall somewhere in between.
Knowing the "replacement level" value at a given position, we can take a player's PER and minutes played and use it to calculate his VA. Chris Paul, for example, has played 2,470 minutes with a PER of 30.06 through Tuesday's games. Since the average replacement level PER for a point guard is 11.0, we take the difference (30.06-11.0) and multiply by his minutes played, returning a product of 47,078.2.
There's one more step. We want VA to mean something, and in this case we want it to be the approximate number of additional points the player has been worth to his team, over the course of the season, relative to a replacement level player. To get to Point A from Point B requires us to divide the result by 67. Yes, 67. Sorry, that's what works. (If you're curious, a point of PER over the course of 2,000 minutes is worth about 30 points to a team, meaning that one point of PER over one minute is worth 1/67th of a point.)
So what we end up with is the formula:
VA = ((Minutes * (PER -- Position Replacement Level)) / 67)
Where Position Replacement Level = 11.5 for power forwards, 11.0 for point guards, 10.6 for centers, and 10.5 for shooting guards and small forwards
LeBron James' VA, as you might have guessed, leads the league. The chart shows the leaders this season; we also have VA posted for the entire league on our stats page starting today.
If you're wondering whether this is further evidence that James should be the league's MVP
yes, it is, pretty much. James leads the league by a fairly wide margin, with Dwyane Wade his only close competitor.
In fact, you'll notice another column running alongside VA called EWA: Estimated Wins Added. EWA is the same idea as VA, except that the result is expressed in terms of wins instead of points. This is helpful if you're trying to figure out the impact of, say, removing LeBron James from the Cavs or Kobe Bryant from the Lakers. Note that the leaderboard for the two categories will always be identical; EWA is just VA divided by 30, since it takes about 30 points over the course of an 82-game season to add another win.
Where it gets more interesting, perhaps, is in looking at the rookie of the year race. A little over a week ago I made the argument that Brook Lopez and Marc Gasol had been more effective -- even though they had played fewer minutes -- than Derrick Rose and O.J. Mayo, and thus that those two should be first and second on rookie of the year ballots.
I wrote the article not knowing the VA of the players in question, but it appears I now have some cover for my position. Lopez, despite playing six minutes per game fewer than Rose, is ahead of the pack in VA, while Gasol is running second.
ESPN.com has added VA and EWA to the PER leaderboard, so you can track the best and worst in each over the season's final weeks, as well as in coming seasons.
In the meantime, keep those e-mails coming. I promise I won't tell anyone if you misspell the four-letter words.
|The spotlight loves LeBron James. So, too, does John Hollinger's new rating system.|
QUICK VA FAQ
Does this replace PER?
Heavens, no. PER describes a player's per-minute quality of play, and in most situations, that's the thing we care about most. But not all situations; sometimes we want to know more about what a player's season-long value was. MVP voting is the most obvious example, but there are several other situations that come to mind.
So is Bruce Bowen the worst player in the league?
No. He has the worst VA because it becomes negative for players with a PER under the replacement level at their position. Of such players, Bowen has played far more minutes than the rest thanks to his on-ball defensive ability (which PER doesn't measure). Those minutes make his rating even more negative than the rest, dragging him to the bottom of the pile. For genuine awfulness, however, one could argue that Adam Morrison (-47.1) and Stephon Marbury (-1.1) have done as much to hurt their teams as anyone else in the league.
What about leadership and intangibles?
This is a tool to guide the logic of award votes, not one to automatically fill out people's ballots. There are any number of valid reasons why a player who ranks high on this list might be more or less valuable in real life, as I've covered ad nauseam in previous missives: individual defense, locker-room presence and leadership, clutch plays, etc.
I tend to see this as a situation in which the burden of proof matters. As in, if you think Kobe is the MVP and LeBron isn't, and LeBron's VA is nearly double Bryant's, then the burden of proof is on a Kobe voter to explain what in his 2008-09 résumé could overcome such a disparity.
Why don't you just call it VORP?
VORP, adopted from baseball, has been used to describe a number of different methodologies in basketball. While they all do the same sort of thing as this one -- differentiate a player's value from that of a replacement-level player -- there are several different means to that end. It seems pretty easy to confuse onlookers if I'm using one method and calling it VORP and other analysts use different ones, so I thought it best for all concerned if this thing had a name of its own.
John Hollinger writes for ESPN Insider. To e-mail him, click here.