Page 1 of 1

STAT-uration - Vol Two: PER

Posted: Wed Mar 12, 2008 12:07 pm
by count55
A reminder of what we're doing:

count55 wrote:With the advent of easy access to computers, fantasy sports, and the internet, there has been an explosion in not only the amount of statistics available, but the sheer number of metrics used to measure the performance of athletes and teams across all sports. It's even reached the point where statistical modeling is used not just as a tool of sports management, but in some cases, as the primary decision-making tool/driver of professional franchises.

I thought it'd be interesting to open a discussion about the value of this turn of events. However, rather than starting an esoteric discussion on the broad nature of the subject, I thought we'd go through different metrics. The idea being that we understand what the metric is, and how it is used. Then we can discuss how valid we think it is, how it should be used, whether it's being used correctly, and how relevant it is to the way we think.

Considering that this is a forum where disagreements arise regularly, maybe it would help for us to come to some conclusions and understanding about the stats we may or should (or shouldn't) use to support our positions.


The second subject will be: John Hollinger's PER

The Definition/Overview wrote:The Player Efficiency Rating (PER) is a rating of a player's per-minute productivity.

To generate it, I created formulas -- which I outlined in tortuous detail in the book "Pro Basketball Forecast" -- that return a value for each of a player's accomplishments. That includes positive accomplishments, such as field goals, free throws, 3-pointers, assists, rebounds, blocks and steals, and negative ones, such as missed shots, turnovers and personal fouls.

Two important things to remember about PER is that it's per-minute and pace-adjusted.

It's a per-minute measure because that allows us to compare, say, T.J. Ford to Jose Calderon, even though there is a disparity in the minutes they played.

I also adjust each player's rating for his team's pace, so that players on a slow-paced team like Detroit aren't penalized just because their team's games have fewer possessions than those of a fast-paced team such as Golden State.

Bear in mind that this rating is not the final, once-and-for-all answer for a player's accomplishments during the season. This is especially true for players -- such as Bruce Bowen and Jason Collins -- who are defensive specialists but don't get many blocks or steals.

What PER can do, however, is summarize a player's statistical accomplishments in a single number. That allows us to unify the disparate data on each player that we try to track in our heads (e.g., Corey Maggette: free-throw machine, good rebounder, decent shooter, poor passer, etc.) so that we can move on to evaluating what might be missing from the stats.

I set the league average in PER to 15.00 every season.

Among players who played at least 500 minutes in 2006-07, the highest rating was Dwyane Wade's 29.04. The lowest was Collins's 3.02.


John Hollinger is a columnist on ESPN. As such, he has a rather large megaphone through which to promulgate this particular metric. If you go to ESPN's NBA page and go to the "More+" menu pulldown, you'll see four (4) separate items with Hollinger's name on it.

So, what do you think of the PER? Does it really indicate that a player is better than another? Or does it just indicate productivity? Is there a difference?

Whaddaya think?

Posted: Wed Mar 12, 2008 2:33 pm
by mizzoupacers
I think PER probably has more value than +/-. But you can still put only so much stock in it. So far as I can tell, it does not measure in any way WHEN players do what they do, or who they do it against. And timing is damned important. If you can be productive against good teams, in important situations, then you are a good NBA player. If you can't, then I really don't care how many points or blocked shots you get against the Sonics in a meaningless game in March.

Posted: Wed Mar 12, 2008 11:50 pm
by Miller4ever
The best way to gauge a player's performance is still to watch carefully, and not just notice when they score or rebound or come up with a defensive play, but what they do without the ball, and how they lock down or chase on defense, because the things that don't show up in any of these stats have anything to do with productivity/good playing.

A few good examples are picks. You don't get a stat for a successful pick, but if it leads to a score, I think it should count as an assist, or half of one. And then there is the save from out of bounds. This is one of the more spectacular hustles out there, but it doesn't get a stat, or a mention from PER, because PER is derived from stats.

He even says it himself that it can't gauge things like that.

Posted: Fri Mar 14, 2008 1:17 am
by floppymoose
PER sucks. I use it when I run out of tp.

Seriously though, it's a boxscore stat. Since the great players are usually excellent at multiple things, some of which are bound to show in boxscore stats, PER will consistently rate the great players highest.

Where it breaks down is in distinguishing lesser players. If you are a role player who is a bigtime help defender, PER blows as a way to rank your contribution.

Posted: Fri Mar 14, 2008 1:49 am
by Charcoal Filtered
Not a big fan of it and have never used it to support an argument. Like that he takes into account the different paces that teams use, but think there are many other variables that have to be included to make a useful statistic.

Posted: Fri Mar 14, 2008 7:33 am
by Miller4ever
Hollinger is number-crazy, but he also keeps things simple, so that people can actually understand him. If someone were more dedicated and didn't have a fan base to appeal to, then a formula would be constructed with many other factors, such as:

-Team's record differential with and without the player
-Saves, picks have some statistical bearing
-Opponent matchup player on both ends of the floor, counted by possession percentage and all that...
-The list goes on and on

But Hollinger took a very crucial, baby step towards gauging players more accurately using math. However, his math right now is not complex enough. What's needed is quantizing of all those intangible factors.