Sports Economists: Always Wrong About Everything

Uh, Mathlete?

Should have paid more attention in Stats 350.

Phil Jackson did say that Rodman was the best athlete to ever play for him.  Not that that means anything to the particular argument that he was more responsible for wins than MJ, but at least he has something going for him.

That Canto III is posted.

And September is just around the corner.

beautiful

This is a problem a lot of people (including myself at times) have. Taking a position contrary to the general consensus with only a small piece of terrible evidence provides an opportunity to look brilliant if you're correct.

It's also pretty easy to weasel your way out of your position if it becomes obvious that you're wrong.

What sucks is that these guys get paid to spew bullshit. What's nice is that it's easily ignored.

All this post needed was some matrix multiplication and linear algebra and I would have been in heaven.

Although really Brian, its easy to see what David Berri's line of reasoning through stats

• Ho: Every expert on Earth is dumb except for me.
• Ha: At least one other expert is not dumb.

Of course, in this view, rejecting Ho does not mean that Ha is necessarily true.  Even if he were wrong, it emphatically does not mean that you are right!

but this:

You have no mechanism. Your correlation is extraordinarily weak. You have just wasted everyone's time

is right on as far as I'm concerned. If you can't identify a mechanism, then you can't provide a causal account of whatever you're looking at. And without causation, you're not going to have an informative (in the sense of being explicable) correlation.

Every time I read something like this it reminds me of how the V2 bombing patterns in Gravity's Rainbow are isomorphic with a map of Slothrop's sexual dalliances.

I once spent a few weeks running regressions on our factories (20 or 30 of them) demand planning  to better attempt forecasting and sent it to my boss as a "look at what I have been spending my extra time on!" thing.  Come to find out regressions can be done with an excel formula (while I actually wrote the formulas myself into the cells). Moral of the story: regressions aren't that complicated.

The problem with all of Berri's outlandish theories is that they are wrong. Not because of old guys who peer into the soul of Andre Ethier and see a ballplayer, but because of other, more careful numbers from people who are looking for things that are true instead of things that are impressive to Malcolm Gladwell.

In fairness to Gladwell, Berri's use of igon values is pretty impressive.

Oooooh Boy, my self esteem just went down a little. Nice, BC. Real nice. /uncomfortable humor

Obvs, this is a topic near and dear to my heart. My beef with Berri is his tact, not his technique. As much as hardcore stats try to create a sterile environment in which to conduct the tests, there is one flaw that is a devil to get rid of: the tester. You can not care what the results are, one way or another. The problem is that it takes so  much effort to do something like this, that's it's hard not to care. In order to test something you must have a hypothesis--your own, precious hypothesis--and that act creates bias in your work that is very easy to lose sight of.

Berri's problem is that he says, basically, "they are wrong AND I, me and me only, am right." A simple dose of humility would go a long way. 

Now, I must go find some humility. Just what I need, another boondoggle...

bomb away!

I have 3 comments.

The dictatorship/repressive gov't means better soccer argument is just the same one that was resolved in the 80s with the end of the Cold War and the USSR. The resolution was, if the gov't is so repressive that it doesn't stick around it hardly matters if it produces better athletes.

The toxo study is a reminder to stick to scientific articles published in peer reviewed journals or at least based on such (and to be skeptical even of those.) Slate ranks about even with the Onion for learning about science.

There was no need to bring Frenchy Francouer into this discussion. He's suffered enough.

I largely agree with Brian's comments, though with a caveat.  And yes, while I did take some statistics courses at UM, I did not enjoy them and would not consider my knowledge any more than adequate, so feel free to point out my idiocy.

Having recently completed Soccernomics, the takeway I got from the book had less to do with absolute mathematical models for the beautiful game and more with some interesting correlations (and some causations) behind why certain regions and teams succeed while others struggle. 

With respect to China and the US, the authors note that while both those countries have vast populations and significant GDP, there are still a myriad of factors that are slowing their progression to the top of the soccer world.  With China, the issue stems from the fact that soccer is relatively knew and that their explosive growth is wealth has been over a pretty short time frame (compared to other established nations), with exploits of those resources lagging behind.  Furthermore, the massive size of the country makes it difficult to centralize these resources and identify the best athletes.  The same goes for India - the book notes that the best soccer countries are those in which a free flow of information occurs, where networks with other soccer-playing countries are strong and the internal communications promote new ideas.  With China and India, they are geograhically isolated countries that are just now learning how to play soccer on a national level.

As for the US, the authors note that the US's approach to developing soccer players (focus on team winning over individual development, lots of practice, reluctance to let young kids turn it into a career and bypass school a la baseball) has stunted development, as has the fact that our best ahtletes do not play soccer (which amongst world powers is quite rare).  American sports like football and basketball are still huge draws, though perhaps that will change as more players are exposed to the sport at younger ages.

Ultimately, the book is more about explaining that England does reasonably well given their size and GDP (but will struggle to be a world power) and identifying the next generation of global powers.  There was some noise in their findings, but saying that the US, China, Germany, and (surprisingly) Iraq will likely become stronger in soccer while older countries like England and Italy will struggle doesn't strike me as particularly foolish, and the data certainly provides some worthwhile context.

As for Berri and statistical models in general, I agree that they are overused and dangerous in the wrong hands, but at the same time there is some truth to his findings.  You look at that list of 1999 win share players (must be based on 1998 stats, since Jordan was out of the league in 1999), and your eyes bug out because Rodman is #1 and Jordan is #6, but 7-8 of those players should probably be on that list.  That Nets team was surprisingly good (43-39) with Jayson Williams averaging a 12.9-13.6, and the next few years they struggled to remain relevant.  Sure, Jordan should have been #1, but I view this more as noise than some fundamental flaw with the model.

I don't know much about Berri, and some of his findings are suspicious, but as much as people criticize certain individuals for torturing the numbers to come out like Moneyball, some of the counter-examples seem skewed in their own ways.  As they say, there are lies, damn lies, and statistics, and I guess I am willing to give a bit more credence to the spirit of the studies even if the execution is wonky.

I love this rebuttal at http://gladwell.typepad.com/gladwellcom/2009/11/pinker-on-what-the-dog-s...

One of his arguments against a person's evidence is that the person supposedly argued that black people are intellectually inferior to white people.  It's a classic ad hominem.

I once got into an extensive debate with an economist about the "environmentalist dogma" taught in schools--such as, "if we want to have trees, we should not cut them down."  He explained that if the demand for trees is high, the supply will rise to meet it--so if we want lots of trees around, we should seek to consume as much fresh-logged timber as possible.

I wish I was kidding.

I've been doing quite a bit of investigation into the correlation between pass rush and pass defense, and running regressions a-plenty.  With no formal training, I've had to continually step back and make sure I'm following the data, and not making the data follow me.  I think, though, that two factors have been crucial in applying these statistical mechanisms to sports statistics:

1) Having a thorough understanding of what's actually happening on the field, both to filter out meaningless correlations, and to dig deeper when "common sense" relationships aren't found, and 

2) Starting with a specific hypothesis that I didn't come up with.  Testing someone else's hypothesis is much easier to do dispassionately than poking sacred cows with numbers and seeing if any pop.

Peace

Ty

To me its an issue of presentation.  If you say "maybe on-field experience for a qb is as important as talent", you deserve more respect than "coaches know nothing about ability". 

Even if they're dead wrong, proposing alternative hypotheses is at least interesting.  Interesting enough to warrant further analysis in some cases. e.g.  Maybe teams who arne't going to make the playoffs should be more inclined to play young QBs even if they aren't top picks.

I wouldn't dismiss the entirety of economic analysis in sports just because of a few over-ambitious dickheads.  Doing so comes off as overdoing it ( not unlike what Berri is accused of).

You touched on the problem here where statisticians view statistics in a vacuum and ignore what is actually happening in the game. I've heard countless arguments about why JD Drew is the best player in baseball and is underpaid based on his OPS. This ignores the fact that he has never had more than 70 RBI and is a terrible situational hitter. Another stat baseball statheads have been known to rag on is hitting with RISP. They all claim that it is a statistically insignificant stat and is just "luck." This ignores factors such as the pitcher having to change from a full windup to a stretch windup, distraction to the pitcher by the baserunners, and the heightened intensity of of the hitter/pitcher in a critical situation.

Stats are great tools but can never capture the full story of sports.

soooooo.... I thought we weren't talking about soccer anymore...

I kid I kid

Go Spain! 

I'm pretty sure Michael Jordan did not play during the 1998-1999 season. His last championship with the bulls was during the 1997-1998 season. He retired (again, sort of) afterward.

good sports economists have the ability to understand that the level of uncertainty in sports can lead to statistical oddities. 

rock on brain, rock on

You are right that he is a hack.  You may have even understated the case.

You are wrong that he has some affiliation with Princeton, or any other above-average institution of learning. 

• He went to Nebraska Wesleyan.
• He got a PhD from Colorado State (i mean no disrespect, but I'll call them average)
• He now teaches at Southern Utah, after stints at Coe and CSU-Bakersfield

http://en.wikipedia.org/wiki/David_Berri

Berri gets cited because he has any school attached to his name.  I am aware of only one professor from a leading school who has addressed Berri's work and it was Steve Pinker in this rebuttal to Malcolm Gladwell's pathetic counter-attack after Pinker beatifully trashed Gladwell's book in this gem.  Pinker correctly concluded that Berri's work was pretty conclusively disproved by . . . bloggers. 

sports is a topic in which any academic must answer to an army of statistics-savvy amateurs, and in this instance, I judged, the bloggers were correct

he really is the best.  Phil's posts are inevitably very charitable to the intentions of the research  to be vivisected, which makes them all the better.

...I've been listening to that book on audio in the car (tables in audiobooks are TEDIOUS) and agree that he is very ad-hoc.  Left by his ad-hoc model to conclude that "Norwegians are the biggest soccer fans on earth" why does he not consider the alternative hypothesis "my model is ad-hoc and stupid".  My chief problem is that he is using the Terrible Hammer of Regression Analysis on variables he picked up in the supermarket.  There is no theoretical reason to include the stuff that he does, he just includes the variables BECAUSE HE HAS THE DATA.

 He seems to have read up on regression on-line, with the same sort of success as that Times Square bomber dude who read up on bombs on the interwebs and ended up only melting his upholstery.

But "never right about anything?"  I like your gusto but not your rhetoric there.  If  you want to play at cutting down their stats you have to do it using their language.  Saying you are 100% about something is like saying you think Obama was born on Mars.  It points you out as Not One of the Clan.   Why not beat them at their own game, rather than make a statement that could be refuted with a single observation of any statistician ever being correct about anything?

I like Berri's book, but not because I agree with stuff in it - instead because it makes you think about the problem a different way.   It's certainly sounder theoretically and mechanically than the ad-hoc Kuper approach.   In the end, though, if the conclusion of the model is whack we need to go out and show why our model is better.  That's the Scientific American Way. 

is awesome.

it's posts like these that make me worry - worry that one BC will become a writer on the national stage at some point, leaving mgoblog behind for a bigger venue (if there is one :)

what an incredible piece of writing. 

the real thing that should worry you: economists are likely doing this in *every* thing they study, not just sports. alas.

When I'm not writing sports posts on a blog that very few people read, my real job is as an economist. And while I take umbrage with some of the broad generalizations, Brian nails the issues with almost any study I've ever done.

There's an old joke that says somebody wanted to know what 2+2 equalled...a mathematician knows it's between 3.9999999 and 4.000001 while an economist will ask, "What do you want the answer to be?"

In 1806 Pfuel had been one of those responsible for the plan of campaign that ended
in Jena and Auerstadt, but he did not see the least proof of the fallibility of his theory
in the disasters of that war. On the contrary, the deviations made from his theory
were, in his opinion, the sole cause of the whole disaster, and with characteristically
gleeful sarcasm he would remark, `There, I said the whole affair would go to the
devil!’ Pfuel was one of those theoreticians who so love their theory that they lose
sight of the theory’s object—its practical application. His love of theory made him
hate everything practical, and he would not listen to it. He was even pleased by
failures, for failures resulting from deviations in practice from the theory only proved
to him the accuracy of his theory.
—Leo Tolstoy