20 Second Timeout is the place to find the best analysis and commentary about the NBA.

Friday, December 24, 2010

The Value of Skepticism

Einstein's Theory of Relativity has led to countless scientific breakthroughs and greatly increased our understanding of the universe, yet physicists are still constantly running experiments to test it and make sure that it provides the most valid description of how gravitational fields work. This is because real science is based on skepticism, doubt and questioning--not credulity, certainty and censorship of opposing ideas.

I used to be an active contributor at APBRMetrics, the website for the "stat guru" wing of the Association of Professional Basketball Research, but I stopped going there after several discussions degenerated into name-calling and after it became apparent that many "stat gurus" have no understanding of (or interest in) the methods of genuine scientific inquiry.

Recently, I stumbled on an article that argued that Gilbert Arenas had once been an elite player. Since I strongly disagree with that idea, I posted a respectful comment and then I wrote a brief article here that linked to the ongoing discussion. The original author of the Arenas piece never really directed addressed my points--let alone attempt to refute them--but some other people offered their opinions and I have been responding to those comments over the past few days. The Basketball Reference.com blog used to contain a link to 20 Second Timeout but, oddly, after I posted my most recent comment there I noticed that the sidebar link to 20 Second Timeout had been removed. Perhaps Neil Paine and his Basketball Reference cohorts don't want to have too many people visiting here and possibly getting the idea that "advanced basketball statistics" have not completely solved all questions regarding individual and team evaluations. That's too bad, because the truth is that if the "stat gurus" refuse to listen to valid criticisms regarding their methodologies and conclusions then it is unlikely that they will ever achieve their goal of better understanding basketball (if, in fact, that is really their goal).

Labels: ,

posted by David Friedman @ 4:22 PM

9 comments

links to this post

9 Comments:

At Saturday, December 25, 2010 8:06:00 PM, Anonymous dsong said...

David, I actually think advanced basketball (and baseball) statistics can be useful sometimes. All you have to do is look at what these sabremetricians say - then go to the window and bet exactly the opposite!

I used to make markets on baseball and basketball futures before the internet gambling ban. My conclusion is that on the whole, most stat geeks have no clue when it comes to predicting champions. If you want good predictions, you're much better off calling gamblers - people who are forced to put their money where their mouth is.

On the other hand, sometimes they come up with hidden nuggets that you can use to your advantage. I used to be a stat geek myself, so I know both sides of the ledger. They can produce useful tools but are horrible predictors for the most part.

 
At Saturday, December 25, 2010 9:34:00 PM, Blogger Codysseus said...

Please keep posting on other websites and linking to them here. It's interesting and informative to read debates such as the one you had at basketball-reference, even when one side is either severely less engaged in the debate (or just doesn't have enough of an argument to keep up). The removal of the link is pretty suspect but don't let it discourage you from posting there. I enjoy detailed critical writing, keep it up.

 
At Sunday, December 26, 2010 12:41:00 AM, Blogger David Friedman said...

Dsong:

For gambling purposes isn't statistical analysis much more useful on the team level than it is regarding the relative value of individual players? I can see the usefulness of data regarding the relative effectiveness of different five man lineups (assuming that the sample sizes are large enough) but I am not at all convinced by much of what the "stat gurus" say in terms of ranking individual players.

 
At Sunday, December 26, 2010 12:53:00 AM, Blogger David Friedman said...

Codysseus:

I was more surprised that the link was there in the first place than that it was removed; sadly, most "stat gurus" seem to be uninterested in responding to logical, well thought out challenges to their views: sometimes I feel like I am trying to reason with members of a cult. In this particular case, it is hilarious that Arenas' fans simply refuse to acknowledge that he might possibly have not been an elite player even though during his best years his team was mediocre and despite the fact that when he missed virtually an entire season his team was fully capable of being mediocre without him! I cited facts and I used large sample size data and the response is either silence or else something on the lines of "but if you squint and look at certain numbers just the right way Arenas was almost as effective offensively as Kobe and LeBron (provided that you ignore that this had nothing to do with winning and that Arenas was a defensive sieve while Kobe and LeBron are All-Defensive First Team members)." Of course, the "stat gurus" leave out the stuff that I put in parentheses but it is funny to read an article in which someone admits to being a big fan of a player and then tries to spend the rest of the time acting like he did some kind of scientific, objective analysis of the player's value. The sad thing is that so many people believe that just citing one or two "advanced stats" makes an argument scientific and airtight.

It should be obvious that I cannot spend a lot of time going to various websites and engaging in intellectual combat for free while my opponents are getting paid and doing everything they can to prevent me from receiving wider exposure--but this Arenas article was so flawed in premise and execution that I could not resist refuting it: you will note that the original author never addressed my most basic contentions and that he eventually stopped responding at all (other than removing the link to 20 Second Timeout).

 
At Sunday, December 26, 2010 1:07:00 AM, Anonymous dsong said...

David:

That's exactly my point. Ranking individual players is really a shell game. And over the long run, it doesn't really matter. Basketball is a team game and the goal is to win games and win championships.

The other goal - of course - is to make money. NBA basketball is a business, after all.

So I think people are really grasping at straws when they're trying to say that Player X is better than Player Y. The question they really should be asking is: let's say that we have Team A - what decisions should they make so that they have the best chance of winning championships and making more money?

As for statistical analysis, probing over past performance is really an inefficient way of predicting future success, no matter how sophisticated the analysis. You will be much more efficient if you simply watch gambling lines and betting markets all the time. This is how I improved my predictive ability. Most successful gamblers I've known operate this way.

As for Arenas, we all know he falls short of "elite" status. There are three guys who have achieved it who are still active - Kobe, Shaq, and Duncan. Everyone else is trying to reach that level. Arguing over whether player X was the 5th best or the 20th best player in the league isn't all that important in my opinion, and ultimately meaningless in the grand scheme of things.

 
At Sunday, December 26, 2010 2:15:00 AM, Anonymous Anonymous said...

Sharp

First things first, Merry Christmas!

I went and looked at the article after you mentioned it here on Dave.

While I don't necessarily think Neil was trying to imply that he felt Arenas was a better basketball player than Bryant, James or Wade, I find it ridiculous that he believes entirely in some statistic which holds that Arenas was a more efficient scorer over the 2 year period referenced. He's drawing a lot of conclusions from numbers that are contradicted by common sense.

I don't know everything that goes into determining offensive efficiency, but I've been watching Gilbert Arenas since he was in college. I've seen him in person a few times (as I've mentioned in comments previously), and actually had a nice conversation with him after his rookie season once. He shoots too much, takes bad shots, and his assists are mostly the result of dominating the ball than facilitating ball movement. He was able to get away with it more earlier on in his career, but its hard to post 28 per game after various injuries have robbed you of your explosiveness.

Either way, simply watching Arenas play indicates pretty clearly that while he is a very talented player, his style isn't the type that generates wins.

Arenas played on a team with two all-star forwards, and pulled of 40-45 win seasons during his prime statistical years. Compare that to Bryant pulling off similar win totals with no fellow all-stars, while also playing against a stiffer pool of competition out west. If Bryant gets robbed of MVP in 06 and 07, then Arenas even being a "dark horse" candidate for doing less with more is a joke. Hell, James made it to the 07 finals with no all-star help.

Someone who really believes in a statistical analysis of the game can rant till they're blue in the face about offensive efficiency or whatever...but the bottom line is that if the Wizards had approached the Lakers/Cavs/Heat in 2006 offering Arenas for Bryant/James/Wade in a straight up deal, they'd have been greeted with a round of laughter and a hangup.

On an unrelated note...what's your take on the Lakers/Heat game? Personally, although ESPN is all over the place suggesting that this means the end of the world for the Lakers, its hard to draw conclusions from one game when there are 50 left to play.

Bryant looked lost out there...didn't know when to assert himself, got stuck in the air with nobody to pass to. Gasol suffered from prolonged minutes at center, forcing him to try to score over lots of long arms rather than using his post savvy to get around them. Jeff Van Gundy ranted about how poor the Lakers' pick and roll defense was, but I felt the bigger issue was bad defensive switches which left Miami with open three pointer after open three pointer after open three pointer. Most of those bad switches were off the ball, and didn't have much to do with the pick and roll.

Either way, if these two teams do wind up somehow meeting in the finals, I'd expect the dynamics of the series to be much different than what we saw today.

 
At Sunday, December 26, 2010 10:14:00 PM, Blogger David Friedman said...

Dsong:

You are right that the larger issue, particularly from the perspective of a team executive, is not whether Player X is better than Player Y but rather which player is most likely to help a particular team win games, playoff series and championships.

If you read my Pantheon series then you know that although I selected the players who I considered to be the 10 best retired players at that time I did not attempt to rank the players within the Pantheon; it is one thing to say that a certain select group of players is elite but another thing entirely to then try to rank those elite players, particularly when they come from such vastly different eras.

I do not regularly engage in discussions about whether someone is the fifth best player or the 20th best player; I tend to categorize players more broadly as "elite," "All-Star/All-Star caliber" and so forth. However, I don't think that people who understand the limitations of player evaluation should just completely abandon the podium, so to speak, to people who use "advanced stats" to promulgate absurd ideas such as saying that Arenas was at one time an "elite" player. Obviously, it is impossible to prove whether Arenas was the 10th best player or the 20th best player at his peak but there is very compelling evidence that he was not a top five player--he was not one of the handful of players who could potentially have been the best player on a championship team.

 
At Sunday, December 26, 2010 10:15:00 PM, Blogger David Friedman said...

Sharp:

As you indicated, whether one uses a broad range of numbers, a scout's eye view or simply common sense there is little support for the thesis that Arenas was ever an elite basketball player.

I provided my take on Heat-Lakers and the other Christmas Day games in my most recent post. I disagree that Bryant seemed lost. Bryant read the defense and made the appropriate plays; he passed the ball to the players who the Heat left open in order to trap him but Bryant's teammates did not step up and make shots. Even worse, the Laker bigs played atrocious defense in screen/roll situations. As Bryant said after the game, the Lakers had a defensive game plan going into the game--and then did the exact opposite of everything that they had practiced!

You are right that if these teams meet in the Finals the dynamics will likely be quite different but the issue for the Lakers--as it has been the past three years--is that Bryant is a lot more focused, driven and aggressive by nature than his teammates are, so he constantly has to push and prod them to reach his level of awareness, discipline and aggressiveness. There are very good reasons that Gasol made the All-Star team once in six seasons prior to joining the Lakers and that Odom has never made the All-Star team; those guys are not as intense or as driven as Bryant. It is odd to hear people wonder why Odom has never been an All-Star; my reply is always to say that such people should pick a year and explain exactly who should have been left off in favor of Odom. Was there ever a season in which Odom was better than Duncan, Nowitzki, Melo, KG (when he was in the West), Amare (ditto), etc.? Odom is a talented player who tends to drift during seasons and even during games, so why should he make the All-Star team ahead of players who are more productive and consistent? While it is true that Odom is playing very well in the first part of this season, he still is not one of the top four forwards in the West.

 
At Monday, January 17, 2011 5:25:00 PM, Blogger $9,000,000,000 Write Off said...

By hiding stats from criticism those sites miss an important winnowing process. The resulting overpopulation means too much numbers and less discernment and understanding. Useful stats are those that take vigorous criticism and survive.

Phil Jackson, Dr. Jack, and Hubie Brown all point to defensive FG% and offensive turnovers as key team indicators of success. I have never heard them mention an individual's statistics.

 

Post a Comment

Links to this post:

Create a Link

<< Home