Recently, there was a post by Dave Berri on the Freakonomics blog (H/T to scraider for the Fanshot) that made a controversial statement - you don't need to tank to be a "good team." Well, I read through it, and it was filled with so much bulls*** that I decided to make a post about it. This economist should stick to analyzing corruption in sumo wrestling or something. Anyways, here's my rant/breakdown of why the article is utterly wrong.
Wins produced is not a good metric
The author of this article, Dave Berri, uses "wins produced" to show that the "best" players in the NBA are not drafted with the first pick. Unfortunately, this is an incredibly flawed stat. Berri would have realized this if he bothered to look at the numbers this "advanced" statistic spits out (edit: I just found out that the Berri is the guy who invented Wins Produced. hahahahahaha of course he did). Landry Fields produces the 8th most wins in the NBA? Ronnie Brewer and Beno Udrih are somehow top-40 players? According to this metric, someone should sign Landry Fields to a max deal! I'm talking to you Otis Smith!
Additionally, there is a problem with his argument - the author attacks a straw man. Berri states that teams tank for the number one pick. He then goes on to cite the number of players who failed after being drafted first, as if that backs up his false argument. Well, no s*** Sherlock, I could have gotten the same answer from a five year old. The point of tanking isn't to get the first pick, it's to get a high pick that might be the first pick. Lots of superstars aren't drafted with the first pick, but they are drafted with a high pick.
The author uses single seasons, and has a selection bias
It's funny how Berri uses single-season data. The good teams, unsurprisingly, are good for many, many years. OKC will probably get 50+ wins for at least 8 years if not more. Dallas, with Dirk Nowitzki, has been good for many years as well. So have the Spurs with Duncan, or the Lakers with Kobe. If a good team wins 55+ games, it's likely they win 55+ games in a future year as well (example: the Chicago Bulls, who had seven 55+ win seasons in a span of eight years). This means that when you look at a 55+ game winning season, you have to look and see if this team was already a 55+ game winner with the same main players and remove it from your data, since it messes up your results. Berri is looking for teams which made the jump, not teams that are consistently good. But when Berri does his "analysis," he does not account for this fact.
In addition, single seasons are a bad way to look at how good a team is. There is variation in a single season, and 55 win teams can fluctuate between 50 and 60 wins. A better analysis would be looking at how each team performed over time, and then relating that to the team's draft position. That would be too much work for Berri, though, and would also hurt his argument
The author withholds and manipulates data
Looking back at single seasons allows Berri to make it appear as though less teams make the "jump" from a bad team to a good team. For example, say it takes three years for one team to get to 55 wins, and four years for another team to get to 55 wins. In Berri's article, these teams would be separated into the "three years to get to 55 wins" and "four years to get to 55 wins" tables, instead of combined into one single table. This makes it appear as if less teams make the jump to 55 wins. He also doesn't reveal any team names, leaving it for the reader to figure out. (Real World Example: Wade's Heat took 2 years to get to 55+ wins, LeBron's Cavs took 4 years to get there)
Moreover, Berri's 55+ win barrier is completely ridiculous. And the focus on wins, in general, is completely ridiculous. What is important isn't the wins, but where they picked in the draft. The whole point of tanking is to get a high pick. What we need to know is the players on those 55+ win teams. My guess is that all of those teams will have a high first round pick leading them. And most likely, that team will have tanked sometime in the past to get that high first round pick.
Of course, this is just a theory, so I decided to back it up. But the article doesn't tell you where it got the data, so I used Basketball-Reference's team season finder to find all the teams with 55+ wins in from the 1976-1977 season to the present. I went through every single season and almost every single 55+ win season corresponded with its main player being a high first round draft choice drafted by the team. What does this mean? You need a high first round draft choice to get 55+ wins. Unless you commit highway robbery (like the Blazers did with the Nets), tanking is the way to go to get a player who is a high pick. Berri's data is useless. He provides no context to his data.
The question is, why doesn't Berri provide any context to his data? Because it would destroy his argument. If he said "the 1987-1988 Boston Celtics were 55+ win team with Larry Bird, but they were also very good previously," people would call him out on his bulls***. Of course they won 55+ games - they had Larry F***ing Bird, Robert Parish and Kevin McHale! Bird, Parish and McHale were high picks! And if the Warriors didn't make that idiotic trade we could have had Parish and McHale (FML)! Berri purposefully holds back information to boost his argument.
In summary, this "Dave Berri" person obviously needs to take some stats classes. He also needs to release the hostages he is holding captive, because otherwise I have no explanation as to how such an uninformed argument got published. And he needs to inform everyone of that lobotomy he had. Thanks for wasting half my day, Berri. I'd spend more time bashing and destroying your poorly reasoned post, but I've wasted enough time.