All--

I'm reading and enjoying a new blog called Skeptical Sports Analysis. It strikes me as exactly the sort of thing that people here like to read, so I'm bringing it to your attention.

Two favorite posts:

The Case for Dennis Rodman

Player Efficiency Ratings: A Bold ESPN Article Gets it Exactly Wrong

All my best,

--Nate

P.S.: I don't write the blog or own it or anything.

Looks okay. That Carmelo article kinda sucks but it's obviously better than ESPN stuff. Most of the NBA posters here feel the same way about Rodman and think he's a legit first-ballot hofer.

Harsh!

rebounding article about rodman is good. "we" know this already, and yes, rodman is a great great player.

actually, there's was a great article showing the diminishing returns of off/def reb. essentially oreb carry weight from lineup to lineup independent of teammates, whereas dreb, not so much. which is what is expected.

the carmelo article is very blah. agree w/ TUT

http://www.countthebasket.com/blog/2...sive-rebounds/

that's the diminishing returns article. guy who wrote it is a frequent apbr poster and works for the rockets. its good stuff

Bobbo, TUT--

Cool. Thanks for responding and for the link to that article.

All my best,

--Nate

Articles like this are out there, yet John Hollinger is still ESPN's stats guy.

Care to explain the problems with the Melo article?

Hollinger has a lot of good stuff though. He is a diehard analytic at heart, but remember, he has to make a lot of his stuff mainstream. The great thing about JH is you get the vibe he really watches a ton of games. He tries to be "more" then just a stat guru.


Sure.

I felt Haberstroh's article was somewhat weak (it was posted it in the offseason thread with a little bit of banter) but Haberstroh actually did an excellent job of showing why Melo is an overrated scorer to the masses. Essentially, his 2pt% and 3pt% are rather low, so his overall TS% isn't that impressive.

When you look at Melo's PPG (3rd in the L last year @ 28.2) people assume he MUST be the 3rd best scorer in the league! (Or something of that nature) But Melo's TS is actually right around average (54.8 last year, 54.4 career, NBA average last year was just under 55) so his actual "skill" is simply shooting a lot.

Obviously this is a much deeper argument (usage v efficiency, turnover rates as pertaining to usage, "double teams" or other teams scheming against bulk shooters, etc.) but the gist is the other guys around him on the PPG list (like Durant or LBJ) are also VERY efficient scorers. So, they are not overrated.



This I agree with. Haberstroh used the blanket statement that Carmelo is overrated to mean Carmelo is not a top 30 or 40 player in the league. The real question should be "how many maxes are there in the league?" On a personal level, I feel Carmelo is very overrated. But I also am 100% positive he's a top 40 player in the league. To me, he's a max guy, and I wouldn't really think twice.

Melo also has a good career APM, and does just fine in other advanced stats). The main criticism was that he doesn't "perform" well according to PER though... This is kinda bizarre; Melo put up a 22.2 PER, which was good for 10th in the league.

The other issue is the blog post is attacking Haberstroh - not Hollinger. PER is Hollinger's baby, not TH. Biggest issue in the blog post is this:

Uhh, Melo was 10th in the L according to PER. I would call that the OPPOSITE of pedestrian.

Actually, PER is one of the most forward-thinking metrics because it incorporates usage. Many others (offensive rating, WP48, and so on) completely ignore burden of offense. It's assumed greater usage = more difficult shots (that's why the marginal shot % doesn't need to be league average to "add" to your PER!).

Melo's TS% is just fine given his usage. He's a top five scorer in the league. Calling him a top 40 player is lol. He's a top 15 player and probably closer to 10 than 15.
The blog post sucked because he wasted time nitpicking a negative of PER that isn't even a big deal when he could have just said,"PER sucks because it doesn't know how to measure defense and loses value as an offensive stat because blocks and steals are included for some ******ed reason."

Edit: To be clear, Haberstroh's initial article was worse.

guy makes a fair point about the context of shots, this was something that has been thrown around a bit here. i'm not sure that 56% ts is strictly greater than say 57% ts on the same usage, which is to say that not all ts% are created equal. players who function as the bailout option will have lower % shots available to them on those attempts, especially if they are creating these opportunities, which may form a decent chunk of their overall attempts. of course this begs the question of whether the shots were taken out of necessity or to the detriment of the offense (in the sense that the player had to take a bailout shot b/c he ****ed up along the way), but it warrants consideration. also the big fail on haberstroh's part was not mentioning fta

Ok, but I think we can all agree that ts is fairly accurate. If one guy has a 60%ts, and another has a 55%ts, you can tell the 60 guys is better. If it's 60/59, look at it on a case to case basis I guess.

uh tut, i wasn't saying he wasn't a top 15 player. (he may not be, but i didnt say that) in fact, i was saying the opposite, he's OBVIOUSLY a top 40 player, and those guys should all "probably" be maxed. so... he's an obvious max.

Bobbo,
I don't have insider so I can't access the initial article. But Skeptical Sports doesn't say Melo has a pedestrian PER, that is Haberstroh's argument.

Also while PER is Hollingers baby there is nothing wrong with criticizing someone for using it incorrectly. They don't need to be the creator to be subject to criticism.

disagree. if they have similar responsibilities as creators yes. if not, then no

i'm enjoying this blog a lot so far, tyty.

Obviously there is some variance to shot selection, but over the course of the season if two guys have the same usage (and same tov%) it's pretty likely that there roles were similar enough where TS% definitely means something imo. Now a difference of one percent might not mean a ton, but I don't think anyone would make a concrete judgement from that anyway.

Ah, my bad.

Yeah, Haberstroh's thesis was god awful and he went about supporting it pretty bad too. I think the blog could have done a better job of arguing against this though as it's essentialy shooting fish in a barrell.

Yeah, maybe you're right. I think the bottom line is that TS%, like pretty much any other basketball, tells you something about a player within the context of his lineups. Even TS% isn't very predicative of what a player would do you if you dropped him in a new lineup overnight.

I'm not sure this is true as anything more than a coincidence. I trash PER with the best of them, but this is an aspect of PER that I've actually defended on here about 5 separate times. The reason that a marginal shot doesn't need to be league average % to add to your PER is because a shot is not a turnover.

I've heard this claim before, but I don't get it. His work doesn't seem like that of a great statistician who has dumbed it down for mainstream consumption, it seems like that of a bad statistician who has still dumbed it down for mainstream consumption.

Bingo (And I say that while still considering one of ESPN's best assets).

I'm not sure roles are this clear-cut, but still, all this proves is that Melo is probably not as good as Durant or James, which no-one thinks anyway.

Not in a vacuum, usage matters.

Obv usage matters, but I don't see how it can be used reliably. Someone can have high usage because they hog shots that would be more efficient for other players to take, or they can have high usage because they are the shooter of last resort. Or some combination of the two. If the first is more accurate, mediocre shooting numbers could be even worse for the team than they seem, and if the 2nd is more accurate, the same numbers could actually be a huge benefit. I just don't see how roles and playing styles map onto that stat in any predictable way.

I actually think Hollinger is a pretty good writer, but as far as analytics go, I think the people who work for him are probably ESPN's best assets. For example, (sorry to non-insiders), see this article:
NBA Free Agency: What to expect from Miami Heat in 2010-2011
From this, I conclude that there is someone named Goldner at ESPN who is probably smarter than Hollinger.