Thursday, May 21, 2015

A Problem with 538's NBA ELO Time Series

UPDATE 22 May: FiveThirtyEight doubles down on this flawed methodology with a follow-up post that makes the erroneous thinking inescapable. Not good.

The ESPN website FiveThirtyEight (disclosure: I wrote a handful of pieces for them in 2014) has just put up a time series of ELO rankings for NBA (and also now defunct) basketball teams. The effort is notable but deeply problematic. Consider that the ELO rankings produce the following:

1949 Chicago Stags: 1577 (pictured above)
2015 New York Knicks: 1256

Raise your hand if you think that the 1949 Stags (a fine team no doubt) would be favored to beat the 2015 New York Knicks!

The problem is that ELO is constrained to an average value of 1500 over time, thus removing the signal of any trends, variability or non-stationarity in the league performance.

ELO is a tool for contemporary comparison, and is not well suited for presentation as a time series because the baseline changes over time. I'd guess over the long term it has changed so that teams get better, on average, as compared with those of the  past. But certainly there is variation in the overall competitive level of the league, which ELO does not capture.

So when 538 says that the 1996 Chicago Bulls were "the best ever there ever was," what this really means is that they were relatively the best team ever, in comparison to the competition that they faced in 1996. That is a bit different than saying they are "the best that ever was." Producing a time series of ELO is thus fundamentally flawed.

In other words, an average ELO team in 2015 at 1500 is not at all comparable to an average ELO team at 1500 in 1949. The 2015 New York Knicks, as awful as there were in 2015, would probably beat the 1949 Chicago Stags by about 100 points.

Thus, rather than reporting absolute ELO scores 538 might have reported anomalies from the long-term average, to present a team's relative ranking, thus removing the need for consideration of any trend. Not as cool as a long-term ranking, but more accurate.

It is a good first effort, but needs some work!

Tuesday, May 19, 2015

Gatlin's Geriatric Sprinting Exceptionalism

The graph above shows sprint times for 100m for 10 top sprinters, by age. The data comes from the IAAF (which hosts a fantastic array of easily accessed data). The sprinters include those with the 9 fastest times ever recorded (< 9.84 seconds) plus Carl Lewis. More sprinters could be easily added.

I was motivated to compile the data by this post today by Ross Tucker at his excellent Science of Sport blog. Tucker discusses the interesting case of Justin Gatlin the US sprinter who at age 33 is running faster than he ever did when younger. Gatlin's circumstances are interesting (right euphemism?) because he served a 4-year ban for doping from 2006 to 2010 for steroids. He is of course not the only athlete on this graph who has been punished for doping.

Tucker explains why Gatlin's circumstances provoke interest:
Since his return in 2010, he has run faster every single year, and claimed six of the seven fastest times in 2014.  He also had the two fastest 200m of the year, including a double at the Brussels Diamond league event where he ran 9.77s and 19.71s on the same evening.

Impressive stuff, and just what the sport needs – a challenger to Usain Bolt (and the other Jamaicans) as we build towards the Rio 2016 Olympics, and Bolt’s quest to claim a third sprint double.

Except, there’s that nagging, not insignificant doping problem around that challenger. Gatlin is the problem that will not go away. . . .  he is a former doper, dominating a historically doped event, while running faster than his previously doped self.
Motivated by Tucker's post I thought I'd look at some data.

The graph shows that these top sprinters pretty uniformly improved their sprint times until ages 25-27.  The average age for career-best time is just over 26 and the average age for 2nd career best time is just under 26. Gatlin's career best (so far) came last week at age 33 and 2nd career best at age 32. At the same meet in Doha that Gatlin won, 39-year old Kim Collins came in 4th at 10.03.

In fact, none of these athletes ran a faster 100m after age 27 than they did at 27 or before (except Richard Thompson who ran faster at 29). Tucker points to a few athletes who had career bests in their 30s. Gatlin, of course, wasn't running (for time) at age 27 since he was serving a doping suspension. We can imagine all sorts of ways to connect the dots between the two red curves in the graph, consistent with a range of possible story lines. Here are a few:
  • He might just be the greatest sprinter in history, in terms of longevity and sustained performance (just look at Usain Bolt's curve), who ran into some really bad luck.
  • He might be, as Tucker and others mention as possible, still benefiting from past doping, suggesting that short-term bans aren't really up to the task.
  • He might be doping today, and has figured out how to evade the testers. 
We will be hearing more about Gatlin and doping to be sure.

A Model for Sharing Licensing Value with College Athletes

On Twitter, Steve Berkowitz (of USA Today @byberkowitz) shared the image above from a presentation given by Doug Allen to the Knight Commission on College Athletics.

The slide shows a proposal for group licensing of players' name and image right. The proposal has a lot in common with an idea that I have aired to model college athlete compensation for their "intellectual property" in much the same way that universities manage IP for professors and other researchers. Allen's proposal suggests that others are also thinking about creative ways to align incentives in a positive way.

Here is my proposal in more detail:
What the NCAA can learn from Bayh-Dole

College sports are facing a crisis. A group of about two dozen current and former college athletes, led by former UCLA basketball star Ed O’Bannon have sued the National Collegiate Athletics Administration. The athletes argue that licensing revenues generated by the NCAA using the images and likenesses of specific players should be shared with those players. In the coming weeks a federal judge will decide whether to certify the case as a class action, which would then bring into the case many thousands of former and current college athletes.

If that were to occur, then the NCAA and universities could be responsible for paying billions of dollars to college athletes. In 2012, the top 5 college athletic conferences collectively received over $1 billion in television revenue for football and the March Madness spring post-season basketball tournament operates under a 14-year, $10.8 billion television agreement. March Madness alone generated more than a billion dollars in TV ad revenue, exceeding that of the National Football League, the National Basketball Association and Major League Baseball. Johnny Manziel, the Texas A&M quarterback who won the Heisman Trophy last year, generated an estimated $37 million in publicity for his university last year.

With the magnitude of the financial stakes, it is only a matter of time before the dam breaks and the notion of the “scholar-athlete” who plays only for school pride and a scholarship becomes a thing of the past. Rather than wait for a court decision, a labor action by high profile athletes or other possible revolutionary changes, the NCAA and universities can get ahead of this issue by paying attention to the lessons of history very close to home.

In 1980 the US Congress passed what the Economist called in 2002 “possibly the most inspired piece of legislation to be enacted in America over the past half-century.” The Bayh-Dole Act changed property rights with respect to the discoveries made in universities as a result of federally funded research. Prior to 1980 the US government retained ownership of the intellectual property associated with discoveries which resulted from federal research and development. Very few of the patents owned by the federal government were being been commercialized, and policy makers sought a way to better capitalize on the billions of dollars in federal R&D taking place at universities.

Under the law, professors and other university researchers who create intellectual property gain a share in its rewards, thereby creating strong incentives both to discover and to commercialize. In the two decades following the passage of Bayh-Dole US universities increased their patents by 1,000% and added an estimated $40 billion annually to the economy. At the same time, the law ensured that technology transfer activities on campus would be closely monitored to ensure that the mission of universities was not compromised.

So what does Bayh-Dole tell us about college athletics? Several years ago, former Senator Birch Bayh explained why the Bayh-Dole Act works: “it aligns the interests of the taxpaying public, the federal government, research universities, their departments, inventors, and private sector developers transforming government supported research into useable products.”

The NCAA and universities should explore aligning the interests of scholarship athletes, university campuses, the NCAA and the sports public with the incredible revenue potential of college sports. Assigning to universities the intellectual property rights of athletes which play under their names while creating a revenue-sharing model with those athletes would meet this need. Just as occuered with respect to the faculty, such an approach would encourage the further generation of revenue from sports, creating a windfall for many college athletic programs, some of which are strapped for cash, and deliver deserved rewards to the scholarship athletes who play the games.

A revenue-sharing model has served college faculty who conduct research and their home universities very well over more than three-decades. Universities should get to work on adopting a similar model for its athletes, before change is forced upon them, perhaps abruptly.

Tuesday, May 5, 2015

German TV Crew Investigating FIFA Arrested and Detained in Qatar

A television crew was arrested, interrogated and had its equipment deleted and destroyed by Qatari authorities while filming a documentary about the 2022 World Cup.

A reporter, cameraman, camera assistant and driver were denied permission to leave the Gulf state for five days after capturing footage of labour camps there for a programme called The Selling of Football: Sepp Blatter and the Power of Fifa.
The reporters were released and their equipment destroyed:
[Florian] Bauer and his team were arrested on March 27, a day after arriving in Qatar, and were held for 14 hours before being released at 4am the following morning.

Admitting he was “scared”, Bauer said: “There were interrogations by people from the intelligence service who said if I didn’t co-operate with them, it would work badly for me.”

The crew was physically unharmed, which could not be said for its equipment.

Bauer added: “Everything was deleted: phone, hard drives. A laptop got destroyed, got opened by I don’t know who.”
The press freedom watchdog Reporters Without Borders ranks Qatar 155th out of 180 countries in terms of press freedom.  Reporters Without Borders argues that with hosting the World Cup comes certain responsibilities:
"Any country with major sporting events such as Qatar seeks the international stage, it is also important that global public be allowed to ask questions." (translated)
For its part, the Qatari government explained that the arrests had nothing to do with FIFA or the World Cup, but occurred to the fact that filming was taking place without required permits. This seems rather circular as the reporters requested permits to film and were ignored. So they proceeded anyway.

Continuing revelations about the Qatar bid to host the World Cup and working conditions within Qatar continue to hound FIFA and the Qatari government. The latest episode adds to the list. The German documentary is set to air next week. FIFA and Qatar can be sure that it will get many more viewers thanks to the arrested journalists. FIFA has apparently not commented.

Monday, May 4, 2015

A Content Analysis of ESPN's Draft Coverage - Evidence of Racial Bias?

At Vice Sports @A_W_Gordon has a fascinating post on the language used to described football players during 15 hours of ESPN's coverage of the 2015 NFL draft.  Content analyses is a powerful tool to reveal focus of attention as well as the key terms used in communication.

Here is a word cloud of terms used by the ESPN hosts only to describe black football players (with the size of the word proportional to its frequency):
And here is a word cloud for those terms used exclusively to describe white players:
Gordon writes:
. . . feel free to stare at this for as long as you want, but let's do a quick breakdown. Only black players were described as: gifted, aggressive, explosive, raw, and freak. Only white players were described as: intelligent, cerebral, fundamentally, overachiever, technician, workmanlike, desire, and brilliant.
In 2014, Deadspin (Gordon and colleagues) did a similar analysis with similar results.

There is also a substantial academic literature on this subject. For instance:
  • Rainville, R. E., & McCormick, E. (1977). Extent of covert racial prejudice in pro football announcers' speech. Journalism & Mass Communication Quarterly, 54(1), 20-26.
  • Tyler Eastman, Andrew C. Billings, S. (2001). Biased voices of sports: Racial and gender stereotyping in college basketball announcing. Howard Journal of Communication, 12(4), 183-201.
  • Billings, A. C. (2004). Depicting the quarterback in black and white: A content analysis of college and professional football broadcast commentary. Howard Journal of Communications, 15(4), 201-210.
And more recently:
  • Angelini, J. R., Billings, A. C., MacArthur, P. J., Bissell, K., & Smith, L. R. (2014). Competing Separately, Medaling Equally: Racial Depictions of Athletes in NBC's Primetime Broadcast of the 2012 London Olympic Games. Howard Journal of Communications, 25(2), 115-133.
  • Schmidt, Anthony, and Kevin Coe. "Old and New Forms of Racial Bias in Mediated Sports Commentary: The Case of the National Football League Draft." Journal of Broadcasting & Electronic Media 58.4 (2014): 655-670.
How we talk about sport tells us something about society beyond sport, and the resulting image is not always so pretty.

Tuesday, April 28, 2015

Kentucky Derby Times and Performance Limits

The graph above comes from a column I had at FiveThirtyEight one year ago. In it I looked at the data on winning times in the Kentucky Derby. Here is the interesting fact which prompted the column:
Since 1949, the time it takes thoroughbreds to run around the 1.25-mile track has averaged 2:02.25, and no winning race time has deviated by more than 3 seconds
Interestingly, horse speeds have plateaued but human speeds have not. Doping? Genetics? Sport science?

The 2014 race, run a few days after my column, played out true to form with California Chrome's winning with a time of 2:03.66.

Have a look at my column for a discussion.

Friday, April 24, 2015

Is the MLS Sustainable?

Stefan Szymanski, professor at the University of Michigan and author of the forthcoming book Money & Soccer, has written a post questioning the sustainability of the business model of Major League Soccer, The post has attracted considerable discussion and debate, as it cuts right to the issue of the long-term prospects for professional soccer in the United States.

Let's cut to Szymanski's bottom line -- after reviewing data and estimates for MLS revenues and expenditures he concludes for 2014:
Overall we arrive at an annual loss of $139 million, or just over $7 million per franchise in 2014. No doubt there are various tax write-offs to soften the blow, especially if losses can be written off against profits in other businesses. But I doubt that would make this a profitable venture overall.
He then surmises that this loss-making situation does not appear likely to change anytime soon, with its impacts softened only by the entrance fees paid to the league for new teams. The bottom line?
You can fund a loss-making enterprise from the entrance fees of new buyers for a while, but without making money, the only reason for doing this would be glory, not profits. Americans constantly tell me that owners of sport franchises in the US will insist on making money. If that really is the case, then I predict that MLS will collapse, and probably sooner rather than later.
His last statement is a conditional, predicated on the desire for owners to make profits. If owners don't care about profits, then of course, Szymanksi is not predicting collapse.

There have been two types of critiques of Szymanski's argument. One is that his numbers are wrong, and MLS is not in fact a "loss-making enterprise." For instance, @kevinmacauley writes at SBNation:
The basic argument of "Giant costs - smaller revenues = big loss, therefore collapse is imminent" only makes sense when you've actually accounted for all revenues. Szymanski hardly got started on that. When someone makes this argument while actually accounting for all revenues, maybe their argument and prediction will be taken seriously.
On his blog Szymanski responded to a similar criticism:
Well, until MLS decides to reveal the books to us we are both speculating. 
A second type of critique focuses on the longer-term prospects of the MLS. Here is an example:
This was always going to be a long term effort with return on investment delayed much longer than your typical stock market play. From an individual team standpoint, losing $5 million dollars per year, 1/20 of that $100 million, the loss is not all that significant when you factor in what these teams, or the league, could be worth 10 years down the road. A $5 million dollar loss per year is peanuts to some of the richest men in sports. Also, TV revenue tripling in 10 years isn’t a bullish thought. The numbers are so low now that it wouldn’t take much to accomplish that. This would more than cover the current losses and that’s before we factor in advertising, sponsorship and other sources of revenue that would naturally increase as well.

If MLS has done anything well, its certainly been the strictness of their cost control. A slow and steady approach has led to an extremely successful league at the ticket window. Once that transfers to television, the sky is the limit.
Arguments about future revenue growth reflect hopes and dreams as much as dollars and cents. The sky may indeed be the limit, and if a profit is to be made then Szymanski's conditional is off in any case. Those critiquing his argument in this manner can just ignore it, and watch the revenues grow year by year.

Here I'll offer a third perspective. Whether MLS is sustainable or not is probably the wrong question. If it is not sustainable, it will collapse -- that is a truism. But if it is sustainable, any realistic assessment of its future would be hard pressed to conclude that it is as "minor league soccer" or, as Simon Kuper recently quipped, as an "elephant's graveyard."

It is just hard to overcome the financial mathematics. Last fall at Sporting Intelligence, I wrote about the staggering wage bill gap between the MLS and the EPL:
On the 2014 [MLS] salary list are 572 players earning total wages of just over $129.5 million. This sounds like a lot of money, and it is. But in comparison to the Premier League, for example, it is small potatoes. The MLS total includes 21 clubs (for 2014 it includes Chivas USA, since folded, and partial squads for NY City FC and Orlando City FC, both being formed). Yet their total wage bills combined equals about the same as the wage bill for Queens Park Rangers ($126 million) in 2012-13, the most recent year’s data available.

In 2012-13, six Premier League clubs each had wages bills well in excess of the entire MLS, with Manchester City topping the list at $376 million (for all club staff). The wage bills for the top 20 clubs in England were almost $2.9 billion combined. For the bottom 20 payers across the Premier League and the Championship it was about $750 million.
The total payroll of the MLS is about 1/15 that of the NHL (data from my new book). Right now the MLS is to global soccer what the Russian professional hockey league is to the NHL -- interesting domestically, but not a threat to the world's top league in any way.

Perhaps there is a way to make MLS more major. Along these lines, Szymanski has previously made an interesting suggestion:
I think the only way around this conundrum is for MLS to become a minor league for Europe and for European teams to buy up MLS franchises. European teams already employ the talent that would make MLS attractive and the big clubs have more talent than they put on the field. Manchester City’s acquisition of the new New York City FC franchise is surely a logical step. Of course, European teams could just loan players to MLS in the same way that they do in Europe, but then there’s the worry that the borrower will not take good care of the asset.
This might be branded as a 2nd-tier Champion's League of sorts, as the clubs with the assets to invest in a minor league farm team would likely overlap significantly with regular Champion's League squads. Such a strategy would also have the advantage of not denying the MLS status as a minor league, but instead, exploiting that fact as a feature.

All this aside, it is great to see Americans debating the MLS, its status and its future. However, in terms of development of soccer talent in the US, I think there are larger issues elsewhere -- like the NCAA, but that will have to await a future post.