A big flaw with PER is how it handles defense and rewarding defense with stats. That's two PER points right here for Blocks and steals, for a player with worse defensive PER according to 82games.com. A similar at best defensive rating, who doesn't effectively cover as many positions as Bryant. Blocks/Steals are tricky and mishandled often by statisticians. Even so Kobe's defense and durability, combined with his ability to assimilate with other players easily due to his repertoire will win out in the end.
This has been my contention with PER as well. I remember a lot of people confused Iverson's high amount of steals for being a good defender. He just gambled a lot, similar to what Wade does now. Wade still doesn't defend the opposing player's best perimeter player either.
I thought the author should have also gone into the playoffs as well. Kobe was unleashed there while also facing nice competition and complimenting his team within the flow of the game. Interesting point as well, I don't think PER adjusts correctly for the type of offense being analyzed. Some offenses are certainly more conducive towards better PER.
Steals and blocks have a definite value in basketball that can't be denied. If you block a shot, you're forcing a miss. If you steal the ball, you're robbing the other team of a productive possession. In addition to the defensive value, there's also an offensive value. Both can lead to more efficient, early-offense possessions. The difficulty is characterizing the other stuff a player does which may hurt his team's defense. A player who gets a steals at the cost of frequently giving up open shots to his man or putting himself out of position can be hurting his team. But it is unfair to make a blanket statement that any player who gets a lot of steals/blocks is doing other things that offset its value. Chuck Hayes is a guy who is near the top of his position in combined steals/blocks/charges. Should he not get any credit for that, because it is assumed that he's playing lazy defense?
i don't think that's a flaw in PER at all. PER doesn't claim to be a good measure of defense. PER claims to summarize a player's stats into a single number and does a good job of doing so. i think the flaw is with the people who want to discredit PER. PER isn't a ranking system. it didn't say that kobe was the 6th best player in the league last year. looking at it as though it's a ranking system and not taking defense and other factors into account is the flaw. this is just a ridiculously dumb statement. kobe is not approximately where he should be in PER rankings? really? he had the 6th best PER. 6th isn't approximately top 5? and then tim duncan isn't listed? surely that list should be expanded to include him and then kobe is exactly where he belongs in the top 6.
You need to read the article next time. Kobe doesn't get assisted on 60% of his buckets, no Duncan was not the "fifth" best PER player if you want to be accurate about it. I think you need to read the part about assisted bucket rates and Defense again. PER is flawed in that is blindly gives out points in those instances, whether accurate or not.
If you contest a shot or alter a shot around the rim you're all forcing a miss, but it's not included in PER. How about a player who blocks a shot out of bounds, the offense gets the ball back and scores on their next attempt? Was that a good defensive play or a net gain of zero?
i read most of it and it didn't seem like it was worth finishing. i've now read the rest. duncan had the 5th highest PER last season. but in general, i don't think PER really is a good comparison post players and perimeter guys. kobe was assisted on a higher percentage of his buckets than lebron, paul, wade, and roy. that obviously isn't the only thing it talks about, but that specific part certainly won't help kobe. using team assist rates instead of individual ones seems strange to me though i guess in a way it helps to adjust for the offensive system being played in. there still would only be a minor change in the numbers if done another way though. i just don't see the complaint or the argument that advanced stats hate kobe. it's a dumb premise for an argument. kobe's PER is in line where he approximately belongs.
PER is trying to summarize the available data in the boxscore as best it can. There are a lot of things that happen on the basketball court which are not recorded in a boxscore, and PER doesn't include those. That's not a "flaw" in the formula, its a limitation of the available data. As per your examples, players who block a lot of shots will also tend to contest or alter more shots around the rim than others. We don't have information for shot contests, but by the boxscore blocked shots tell you more about it than anything else. As for blocking a shot out of bounds, yes that's a positive defensive play. Just as deflecting a ball out of bounds is a positive defensive play. Not as good as a block+rebound or steal, but it still helps.
Actually the net gain could be significant. Howard isn't a better offensive player either. "tim duncan isn't listed? surely that list should be expanded to include him and then kobe is exactly where he belongs in the top 6." Lol "exactly" where he belongs. Next time, actually read this boring article. A lot of boring material is educational, just walk it off next time. Uh, yes it will. See above. "A minor change", not really. I've seen durvasa calculate a net gain of 2-3 (subtracting a point from a big, adding one to a perimeter player) in a very old discussion about 2006 PER rankings. Certainly important to know. PER does try to calculate defense, that previous comment doesn't really make sense either.
howard isn't a better offensive player. not sure how that is relevant in regards to this. kobe belongs in the top 6. i assume you wouldn't dispute that. he had the 6th best PER. the guy really listed 5 players and said that 4 of them were in the top 4 of PER but kobe wasn't even close to where he should be. kobe was 6th in PER which is pretty close to where he should be. i did read the article and didn't call it boring, just not worth finishing. see below? i've already stated that i don't think PER is really an accurate comparison between post players and perimeter guys. yes, it uses the defensive stats that are provided as part of the formula but it doesn't pretend that it gives an accurate measure of overall defense. kobe is obviously a better defender than wade. so you can look at their PER and the gap between the two as players obviously is not as wide as the gap between their two PERs. i think the comment makes sense. kobe had the 6th highest PER. that is approximately where he belongs as far as his rank among players in the league. kobe isn't the 6th best player in the league(he's better than that obviously), but it's close enough to not make a huge deal out of it.
I think part of the article was quite interesting. Great books have a lot of superfluous material. Not many places where you can find the PER difference in blocks and steals to a precise figure. That too. ;] In a critique of PER? Very. Your comment was a bit confusing. Nice, agreed. But it clearly tries to quantify an aspect of defense. The article pointed this out in more detail, which is useful information. I would consider top 6 a bit offensive, but that's became I am competitive. I personally would be upset if I had Kobe's resume and my PER was "24.2".
Time to point out the elephant in the room. At the very end of the article, the author hits the real issue: Kobe's FG% is nothing special when compared to the other top PER performers. PER is about efficiency, so of course FG% is going to carry weight! Arguing over the value of efficiency is more philosophical than statistical.
Initially it's a positive play, but if your man scores on you the following play because you didn't make a defensive stop then it's net gain of zero. If I block a player's shot right back into a player's hands which puts me out of position and he scores right after that's bad defense in my opinion. However, PER rewards me for getting a block despite the player I'm responsible for scoring. Why create a formula with limited data and then pander it off as something relevant to comparing NBA players in real life? That's like taking a poll and using a biased sample of voters to skew the results you want to your agenda. PER is fantastic if you like playing fantasy basketball in comparing one guy to the next and trying to find "sleepers" but in real life it's pretty much useless in comparing one player to the next. Hollinger or some other statistician needs to create a better model. I'm sure one exists, but doesn't have the fanfare and media attention because the person who created doesn't work for ESPN. Currently Marreese Speights is ranked #4 in PER this season. The casual fan probably never heard of him. Is he really the 4th most efficient player in the league? Common sense answer should be no.
