ITT: Discuss the merits of using PER/Box score production to

[jwilde86]

ITT: Discuss the merits of using PER/Box score production to rank NBA players...

Early days yet, but we're approaching something resembling a decent sample size.

http://insider.espn.go.com/nba/hollinger/statistics

^^ Is this a good ranking of the NBA's players?

If so, what happened to studs like Lebron, Kevin Durant, Duncan and Wade in the span of a year? And when did Al Horford become a top 5 NBA player?

If this is NOT a good way to rank NBA players:

Was Lebron James better than Kobe/the best player in the NBA the past two years? If so, based on what? (Kobe won the rings and in this hypothetical PER/stats are not good ways of ranking players) If not, how should players be ranked?

[kabstah]

Lebron, Durant, Duncan, and Wade aren't playing as well as they were last year. Duncan because he's 34 years old with a lot of mileage, and the other 3 because they're off to a slower than usual start. Futhermore, I don't think anyone reasonable, including Hollinger himself, views PER as the absolute bible on ranking NBA players.

That said, there was a PER difference of over 9 points last season between Kobe and Lebron. That's bigger than the gap between CP3 (#1 in PER) and Andre Miller (#50) this year. Would anyone really argue that Andre Miller has been better than CP3 this year? Even if you disregard PER completely, Lebron had the edge in literally almost every statistic available -- raw, per minute adjusted, advanced (TS%, WS, PER) etc.

The better question is, what argument does Kobe have over Lebron? Because his team won the championship? To believe that necessitates believing that the best player's team wins the championship, (and as a corollary, that in any given playoff series), 100% of the time with no exceptions.

[Doctor MJ]

-It's a decent first pass analysis. I'd always do deeper analysis though.

-What happened to LeBron & Wade? Wade & LeBron respectively.

-What happened to Durant? I'm not sure what the cause is, but he clearly isn't scoring like he did last year. Possibly it's related to Westbrook's new role? Dunno.

-What happened to Duncan? Age.

-Since when did Horford become top 5? Ha, good question. This is an example where I have a tough time following PER. Obviously part of the deal here is that PER is a per possession played stat, and Horford isn't playing as much as other league leaders.

-I think LeBron was clearly superior in '08-09. I think last year was debatable. I do weight PER type stats, I also weight in +/-. Beyond that I'm really hesitant to knock a guy simply because his team didn't win the title. I watched the Cavs lose in '08-09, and I didn't see what LeBron was supposed to have done wrong.

A common criticism of LeBron after those losses and the current Heat struggles is that he has to dominate the ball too much, and thus can't make proper use of his teammates. There's a decent point in there, but the Cavs offense was really quite effective. If you make your offense effective, I don't really care how you do it.

[Dr Positivity]

I haven't watched Atlanta often this year, but I'd venture their new motion offense is boosting Horford's assist numbers and making it seem like he's Shaq/Duncan/Hakeem-ing it as a passer out of the post. 3.2apg on a bottom 5 pace team is really high for a C. I don't know why a 19/10 .63 TS% guy per 36 would be 25 PER instead of like 20 otherwise, I mean those are great stats, but not MVP numbers. He's not even blocking more than 1 shot a game.

[mysticbb]

Who cares about the blocked shots per game? Seriously, there isn't a strong correlation between blocked shots and defensive efficiency anyway. Blocked shots are one of the more overrated stats in the boxscore.

Al Horford can back that up with pretty nice +/- numbers. His adjusted +/- is among the league leaders. If a player can produce huge boxscore numbers and +/- numbers, he most certainly is doing something right on the court. Horford has great hands for a big, his passing abilities are very good, he can handle the ball well and can finish around the basket. His offensive play is really efficient. He is a great rebounder, because he isn't just going for the ball, but also focussing on boxing out. He is in fact a better rebounder in a team concept than Kevin Love. Sometimes the raw numbers aren't telling you much. I like Horford since the second half of the 2008/09 season. He improved a lot and is able to play basketball in a successful way. He is much more valuable to the Hawks than Joe Johnson.

Anyway, I see a lot of value in boxscore metrics as long as those numbers are put into context. PER is production on a minutes basis while adjusted to pace. That doesn't tells you anything about the skill set of a player, or how he would match up against certain teams. In average those numbers can show a high correlation to winning. Usually a player who scores a lot even on a lesser efficiency will get overrated. A boxscore is biased, there are more informations about offense than defense in it. Thus players with a bigger defensive impact will get underrated.

http://bbmetrics.wordpress.com/player-ratings/

That's my own rating, which has a 0.92 linear correlation coefficient to winning (regression analysis made with data for the last 25 years). It is similar in the end to PER, but does a better job in terms of defense.

[Vinsanity420]

PER basically tells you who the box score MVP is... does it really mean much? Nah, not really. I have no real idea what actually happened in the game/how well a player played just by looking at the boxscore. I would take PER with a grain of salt.

But if an all time offensive great places relatively low on PER, I would look into that player some, maybe look at offensive W/S, O Rating, whatever the heck I can get my hands on. Take a look at all time Top 20 PER Greats

1. MJ
2. LeBron
3. Shaq
4. Admiral
5. Wilt
6. Wade
7. Pettit
8. Duncan
9. Johnston
10. Duncan
11. Kareem
12. Magic
13. Karl Malone
14. Dirk
15. Hakeem
16. Dr J
17. KG
18. Kobe
19. Bird
20. Oscar

A very messed up order of players - but no denying that they're just Great, great offensive players, though Johnston kinda stands out... Don't know enough about him as a player to say why he ended up there.

[NYK Dolemite]

I don't think that hoops is naturally given to advanced metrics because of the collaborative nature of each play in a game (other than shooting free throws). However, I like PER as a quick and dirty tool to try and at least eliminate the system biases that some players enjoy when counting stats and per game averages are put out there as markers.

It's not perfect but it's good to have a tool to evaluate basic offensive proficiency in a somewhat neutralized way.

I do think it is interesting to see some of the angry responses to people citing to PER and VA and such (if stop calling them VORP and Win Shares, my arguments might be better!) -- but it's reminding me of where we were five years ago in the Tools Guy v. Sabermagician wars in baseball.

PER is more useful once people stop thinking that the point of PER is to declare who the 'best' basketball player is in the NBA. It's just a measurement of accomplishment, just like PPG game. It just has more variables involved.