clock menu more-arrow no yes mobile

Filed under:

Statistical analysis or numerical hackery?

If you buy something from an SB Nation link, Vox Media may earn a commission. See our ethics statement.

I'm a big proponent of using statistics to look at sports in novel and interesting ways. I was reading Rob Neyer, Baseball Prospectus and Ron Shandler years before Michael Lewis wrote Moneyball, and quite simply, it changed the way I look at baseball.

I'll admit it, though, I'm far from the bleeding edge when it comes to applying heavy statistical analysis to basketball. I don't visit 82games nearly enough, but that's going to change, because every time I do check it out I'm consistently impressed. Plus, I was happy to discover the excellent "game flows" over at PopcornMachine this past weekend (despite the fact that the first one I really analyzed was Saturday's train wreck in Dallas).

But even though I'm not extremely well-versed in Advanced Basketball Statistics 401, I have a pretty good handle on Watching Basketball 101 and Reading Box Scores 101, and that's all I really need to debunk this article about Darko Milicic.

More (perhaps too much) after the jump:

I never heard of the site, Hoops Analyst, or the author Ed Weiland, but somehow, someway I stumbled across the article on Monday. The intro caught my eye:

Thanks to an e-mailer, I decided to take a look at Darko Milicic and the path his career might take.

But I quickly lost interest once the author revealed just how much prejudice he brought into his analysis:

We all know Darko’s saga. He was drafted before all-star Dwayne Wade. He was drafted before solid performers Carmelo Anthony, Chris Bosh, Kirk Hinrich, Luke Ridnour and Josh Howard. Several other players from the 2003 draft who haven’t broken through as starters yet, including Mike Sweetney, Nick Collison, Mikael Pietrus, Kendrick Perkins and Leandro Barbosa appear to have brighter futures than Darko. It goes without saying that any one of these players could have been the difference for Detroit in last years’ finals. It’s amazing to me that NBA scouts, who usually do an excellent job of sorting out the players from the pretenders on draft day, missed so badly on this guy.

Yawn. Just another guy writing off Darko (and placing an odd amount of trust in NBA scouts). I saw tables and tables of data below the intro, but the intro already bored me too much to bother looking at them closely. On Tuesday, though, I saw that Motoring linked to the article with this description:

This is a nice article on our favourite Serbian at… the usual site that specializes in articles on individual people. I think it’s a pretty good read and I love the funny block numbers. I’m not the only one who still has hope for him, apparently.

OK, maybe I was wrong. Maybe it was a fair-handed article after all. So I went back to it, determined to finish it to the end. It was a bit of a struggle, but I'm glad I did. Because it assured me that my initial instincts were correct.

It's not just that I disagree with the conclusion --

To recap, if Darko progresses at the average rate of a teenage forward or center who entered the league in the past 10 years, it’s unlikely he’ll continue as an NBA player. He simply won’t be good enough. If he improves in each category at the highest level established by players who entered the league as teenagers, he’ll become a solid part-time center. Even if Darko exceeds this and becomes a solid starter logging 2500+ minutes per season, it’s extremely unlikely he’ll ever be so good that he was worth passing over Carmelo Anthony or Dwayne Wade for.

-- it's that I completely disagree with the premise the whole article is built on. The author compares Darko's production over his first 400 minutes to the production of other players that were drafted as teenagers. Then he looks at the rate of how those other players improved, somehow construing that to be some sort of baseline for how Darko will in turn develop.

OK, first, my problem with the stats: sample size is everything, and 400 minutes is the equivalent of only 8 1/3 full games. You can't use that as any sort of foundation for what a guy is going to do in 2006-07, which, incredibly enough, the author does.

Second, and I'm surprised this point was never made, the Pistons were already a championship contender when Darko was drafted. The kid didn't play a single meaningful minute in his career until he started the final two games of last season. Under Larry Brown, Darko only played the final few minutes of blowout wins and blowout losses so it's not a big surprise that was never very confident with the ball in his hands. That, more than anything else, explains his .262 and .329 FG% the past two years.

Third, the projected constraints on Darko's expected rate of improvement were arbitrarily calculated. Admittedly, in fact:

Here’s how this group performed immediately following the first 400 minutes. The minutes vary, because I broke them off wherever it was convenient, like at the end of a season.

Not exactly the most precise way of going about things, I must say.The article is misguided at best, and a convuluted example of someone mangling statistics to reveal an opinion they already had at worst.

Call me stubborn, but I don't care what anyone says about Darko -- he's always going to be a completely unique case. He was the youngest player drafted in NBA history, and unlike any other lottery pick, he was drafted by a team that didn't need a single significant minute out of him to win an NBA championship. Plus, there are a host of other factors that shouldn't be lightly dismissed (ie, the language barrier, the culture shock of leaving Serbia and Montenegro for the US, etc) that most other lottery picks haven't had to deal with.

But even if you're going to stick with the numbers and ignore everything else, you still can't base your entire argument on 400 paltry minutes. By this "study," Darko's downfall was his poor field goal percentage as a rookie and second-year player, which was apparently so bad that it's unlikely to improve beyond .410 by 2007 -- despite the fact that we're basing this on less than 150 attempts, almost all of which came in garbage time. It's this kind of weird logic that results in leaps of faith like this:

Darko will probably never be a good scorer.

. . . and this:

He’ll become the defense/rebounding part of a rotation at center/PF. This is probably his ceiling.

Blah, blah, blah, blah. . . if you've made it this far, I commend you, because even I'm growing weary of this topic, and I admit my thoughts aren't organized nearly as well as they should be. But real quick, take a gander at Darko's stats this year. He'd not lighting up the world, but entering Wednesday's game he's shooting .467 through seven games. That's not enough playing time to prove anything, but it doesn't bode well for this author's theory. I have no clue at all what Darko's future looks like, but I do know that his teammates are his biggest fans, sticking up for him in the press and praising him when they don't have to. And since his teammates are the ones that have actually seen him play on a regular basis, I'm going to defer to their judgement until I've seen enough with my own eyes to discount them.

You can't get away with this type of "analysis" in baseball, because there are too many people who know how to correctly apply statistical analysis that will call you out on it. This type of career-trend analysis in basketball is relatively new, and without knowing anything about the author or his site, I'm going to go out on a limb that it's being done by an amateur simply trying to make a name for himself. Which is fine, I suppose. It goaded me into devoting far too much time than I ever intended (or wanted). But it's self-serving and misleading, and one of the reasons why a lot of traditional sports fans don't trust "all that numbers stuff," which in itself is a shame. But I'm also curious about just how much value you can really get when you dig too far into the numbers. The really good players and the really bad players stick out like a sore thumb on a basketball court, and there's only so much that you need to rely on statistics to tell you that you might not be able to pick up with your own eyes.

(Why did Motoring post the link in the first place? I have no clue. Maybe they didn't actually read it all the way through, or maybe they were fishing for Pistons content, since half the stuff posted over there is anything but... Zing!)

Darko's Future [Hoops Analyst]