That is what bugs me about those who go on rants against PER. They seem to think that it is supposed to be used as an absolute player ranking metric. I don’t know of one PER user who claims that. It is comparable to when someone claims that number of championships won is part of how to assess the greatness of a player and somebody else retorts “So Horry is a top 5 all-time player?” Of course not. Nobody is claiming that. They just want to include number of rings in the discussion, not have it be the entire discussion.

I like PER because it is very good at what it is there for. It rolls up all the box score contribution of a player into one number. It has its flaws, but there is not a better stat that does the same thing. Again, EFF is the same stat. It’s just the total one rather than the per possession one. That means they have different functions. If you want to know whose box score production has contributed more to his team, look at EFF. If you want to know who has been more valuable to have on the floor at any given time (again in terms of box score production), look at PER.

And of course, a strong PER over not many minutes means absolutely nothing. The margin of error is just far too large. For instance, if a guy played one possession all year and hit a shot on that one possession, he would have far and away the best PER in the league.

And yeah, I’m not saying that the value assigned to any stat by EFF/PER is some sort of perfect metric. I’m just saying it isn’t some guy saying, “I think this stat should be worth x.” It is based on a statistical estimation of the average value of such a stat. So a rebound one season may actually be worth more than a rebound in another season depending on factors like what percentage of the boards are taken in by the offense or defense and how effectively teams score that year.

Many people have complained that rebounding is undervalued and scoring over valued. I would point out that scoring inefficiently is barely valued at all and can actually have negative value. Scoring efficiently is the fastest way to bring up one’s EFF/PER though. The arguments are probably valid but I think it makes sense because at the end of the day,t he winning team is the one with more points, not the one with more boards.

Also, offensive rebounds actually tend to be quite valuable to EFF/PER. Defensive rebounds are not valued very highly, but again, I think that makes sense. Most defensive rebounds, if the player who collected it had not done so, it still would have gone to a member of his team. That means the fact that he grabbed that rebound was not actually very valuable at all. Of course, some players get a higher percentage of boards that would have gone to teammates and some get a higher percentage of those that would have gone to opponents. But since the box score doesn’t show which are which, PER/EFF obviously can’t account for that.

I would say that what PER/EFF most undervalues is actually blocks. Because, to the best of my understanding, it doesn’t account for the fact that by and large, shots that get blocked are close to the rim and have a significantly better chance of going in (if not blocked) than average shots.

Anyway, that is hardly the point, because a “massive” difference int he number of shot attempts for someone like Monroe would be maybe 4 more a game. Even if shooting is considerably more tiring than setting screens, getting open, running plays, boxing out for boards, and all the other things that must be done on offense regardless of who is shooting, there is no way I will buy that putting up 3 or 4 more shots per game would reduce the minutes Monroe could handle by anything significant (2 at max).

But the moral of the story is that just because you don’t understand how something is calculated doesn’t mean it isn’t done objectively.

While PER and EFF are more confusing than ppg and pp36, the comparison is still totally apt because the confusion does not stem from possession adjusting EFF into PER, it comes from starting with a conglomerate stat like EFF. In fact, the adjustment to PER is even better than the adjustment to per 36 minutes. Because it doesn’t just account for court time but for total opportunities.

But yeah, assuming a list of the top PER performers will read out as the top players in the league is like assuming a list of points per 36 min leaders will read out as the best scorers in the league. By and large, it will be close, but it will have some egregious errors caused by small samples. You just have to ignore the players that don’t play much.