PER: Explaining the Howards and Bowens
Updated: October 19, 2006, 2:35 PM ETBy John Hollinger | ESPN Insider
In the wake of this week's posting of Projected PER stats for every player in the league, my mailbox runneth over.
Dwight Howard No. 53? Eddy Curry ahead of Jason Kidd? Bruce Bowen No. 315? And who the heck is Chuck Hayes? Clearly, some folks are confused, so let me try to explain as much as I can in one neat little package. First, let's make sure we're clear: Those rankings you saw were not me saying that Curry is better than Kidd, or Hayes is better than Howard. Thousands of straw men have suffered horrible deaths in the past 48 hours based on this false premise. Here's what the ratings are: A projection of each player's per-minute statistical production this season. Per-minute, for example, Curry is a monster. He just can't play more than 20 of those minutes on a lot of nights because of fouls and conditioning. So in any kind of per-minute rating, he's going to grade out much better than he would if you were just listing players by their overall value. Unfortunately, three kinds of players give the projections problems -- very young ones, very old ones, and ones who haven't played many career minutes. This is where Howard and Hayes come into play. The projections are built by looking at comparable players at the same age and how their stats changed in the following season. For players in most age brackets, this is extremely reliable, but there have been so few players to turn pro out of high school in the past two decades that we have a very small sample to work with for players like Howard and Josh Smith. As luck would have it, one of those players was Darius Miles, who did a barrel-over-Niagara-Falls routine at the same age. As one of just a handful of players used in the projections for both Howard and Smith, he's enough of a millstone that the overall projection foresees a decline in PER. There were two ways I could handle this: (1) "Cheat" by throwing Miles out of my database and having more defensible projections for Howard and Smith or (2) include the projection and explain what happened in the player comments. I opted for (2), not realizing that one consequence of shifting this information from a book format to an online format was that a great many readers would see Howard listed as "No. 53" long before they every got to the explanation of how it happened. (And for the record, I expect both Howard and Smith to blow those projections out of the water.) Making matters worse, Howard is listed well behind somebody named Chuck Hayes, an undrafted free agent who played for Houston and just barely met my 500-minute threshold for being included in the projections. It must be said that Hayes played phenomenally well in his limited playing time -- he finished with the No. 6 Rebound Rate in all of basketball and shot 56.2 percent. I have yet to meet anybody who expects him to keep playing at this level, but right now it's all the information the projections have to go on, so his projected numbers for next year come out looking very solid as well. Keep in mind also that we're not talking about career value here; we're talking about projected stats for next season only. One reader asked why I gave Chris Paul more "respect" than Allen Iverson, but respect had nothing to do with it -- it's the historical trend that point guards in their second season tend to improve and small players in their 30s tend to decline. Either player could be an exception, but based on analysis of similar players, this is my best estimate for their performance this coming season. The projections also don't include defense beyond blocks, steals and fouls, partly because the league has opted to make this area a black hole in terms of stats and so we have nothing to work with. The San Antonio Express-News today wondered why only a handful of players were projected with a lower PER than Bowen -- well, he doesn't score, rebound, pass, block shots or get steals, so where were they thinking he'd rank? Statistically, he really is near the very bottom of the league -- he's just so good defensively that it offsets his lack of numbers. I've noted in the player comments for the Hayeses and Howards of the world where I disagree with the projections, but unfortunately these infrequent examples seem to be dominating the discussion. In the vast majority of cases, we end up with projected numbers that make perfect sense. Curry, for instance, projects to average 22.0 points per 40 minutes and shoot 56.5 percent from the floor; since he averaged 21.0 and 56.3 percent last year, this shouldn't be controversial. The fact this gives him a better projected PER than Kidd may strike some folks as odd based on the two players' reputations, but if he outrated Kidd on this metric this season it wouldn't shock me at all.
Ron Turenne/NBAE via Getty ImagesNo, John Hollinger doesn't think Curry is better than Kidd.
• For the complete list of 2005-06 ratings and 2006-07 projected ratings, click here.
John Hollinger, author of "Pro Basketball Forecast 2005-06," writes for ESPN Insider.