By: Edward Egros

texas

A Review of SXSW

Pasted Graphic
For one weekend in Austin, while others attended festivals, concerts and technology symposiums, I wanted to explore the sports and media panel discussions at SXSW, a conference billing itself that celebrates the convergence of the interactive, film and music industries. First, to be a panelist at SXSW, you must master the art of the humble brag. For instance, to paraphrase a filmmaker's list of his two main goals for any of his products, he said that he must love his work and his audience must love his work.

Pause to be inspired.

A couple of panel hosts reiterated their resumes aloud without using this background for an educated question.

Pause to emote.

One of the better tidbits of advice I ever received as a writer is to know your audience. While SXSW attracts a diverse group of scholars, professionals and fun-lovers, often times this gallery wants to learn how to become the person they are watching, or at least morph into a unique iteration. If keeping trade secrets serves as the highest priority, at least teach us something that will help an aspiring audience. A disappointing talk featured a panel promoting "
The Weekly", a television endeavor from the New York Times committed to long-form storytelling. Instead of previewing at least one specific story without a fancy montage, the conversation felt more like an academic presentation, philosophizing about why this format "will" work. I did want to learn more about the program, but the panel felt so much like an infomercial, Ron Popeil should have moderated. I wished for something with takeaways for my own work as a journalist, not a payment installment plan.

Often times panels with star power offered fewer lessons about the ways of the world than those who let their experience speak for themselves. One exception was NBA champion Chris Bosh, an advocate of analytics and
an elite karaoke singer. It took losing to the Dallas Mavericks in 2011 and signing Shane Battier to make Bosh a believer in the numbers, so much so that he rightfully credits the adjusted approach with his 2012 NBA Championship. Alongside Spurs general manager R.C. Buford, the panel devoted to the future of basketball discussed the importance of the three-point shot, and how big men are taught to handle the ball like a guard and be unafraid to launch from deep.

Perhaps my favorite panel featured a discussion on how data has changed the NFL. In particular,
Sarah Bailey (analyst with the Los Angeles Rams) and Namita Nandakumar (analyst with the Philadelphia Eagles and noted pugilist) discussed how they make the most of a small analytics staff and why any quantitative strides should be celebrated, not interpreted as a "glass half-empty" endeavor. For instance, while quants wish coaches would go for it on 4th down more often, Nandakumar advises to at least take satisfaction coaches are getting smarter about it. A sobering reality to consider is how few jobs are available to data scientists in pro football, so self-promotion and engaging research projects in other sports are vital in this competitive industry. For nuanced advice for data scientists, Bailey offers this advice: when working with large datasets, take a subset, crunch the numbers in the R programming language, and if you are satisfied with the results, take a larger dataset and use the Python programming language.

The last lecture I attended featured Paul Bracewell, Managing Director of DOT Loves Data, discussing how to use machine learning to rate athletes in a variety of sports, most notably international events like rugby and cricket. Some of the better instruction he offered involved how to discuss analytics with coaches and players: "When predictivity is used as a benchmark, the model needs to generate supporting output to explain any departure from the predicted results". In other words, explain why a model works or doesn't work in a specific situation. Transparency and meaning to build trust are of the utmost concern.

I wrapped up my time recording an episode of "
Outside the Box" with our usual EPlay crew. It featured a spirited conversation as to who is responsible for preventing young players from a catch-and-shoot approach to basketball that some believe analytic enthusiasts have espoused. Trust me, you want to listen to this one. Until then, my biggest takeaway from my first SXSW is the opportunities to learn and share ideas exist, but they often remain hidden. Until I revisit it all, I will take a break and watch a video of me petting a robot puppy at one of the innovation exhibits.

No Range for the Texas Rangers

IMG_5937It's hard not to catch shortstop Elvis Andrus smiling these days. His Texas Rangers go into the postseason with home-field advantage all the way through the World Series—while finishing one victory shy of a franchise record for most wins in a season—and boasting the most wins at home in the American League. Elvis himself finished the regular-season as a .302/.362/.439 hitter. And yet, a few sabermetricians have spoken out, saying not only shouldn't the Rangers be one of the favorites to win the World Series, their success is virtually fraudulent.

It involves
Pythagorean Expectation. This is the often-cited formula baseball guru Bill James invented to estimate how many wins a team "should" have based upon how many runs they scored and allowed. Since it became commonplace, the formula has worked quite well explaining why teams are thriving and struggling. Even this season, the formula explains all but a handful of wins or losses for every MLB team. The one team the formula has done the poorest job with, is the Texas Rangers.

For much of the season, this team's Pythagorean W-L hovered around .500. The Rangers finished 13 games above what was expected, at 95-67. Why? The Rangers were 36-11 in one-run games (the .766 winning percentage is a record in modern baseball). They were also 18-24 in games decided by 5+ runs. In other words, the Rangers won a lot of close games and lost a lot of blowouts.

This large of a discrepancy is unprecedented in the last decade for the Rangers:

Pasted Graphic

The Rangers have performed roughly what was expected, given their runs scored and allowed. But the last two years this team has over-performed. It might be a coincidence those were the two years Jeff Banister has been the manager of the Rangers, but maybe not. Banister has a history of evaluating players and looking at skills during blowouts. He is certainly not the only manager to have this approach, but it is possible he takes it to the next level. Two years is not sufficient data to make such a conclusion, but it is a noteworthy trend to consider.

So how accurate is this formula when predicting if the Rangers will win the World Series? Not very. Since 1969,
11 teams out of 47 had the best Pythagorean Expected record and went on to win the World Series. In fact, the likelihood has decreased since the postseason expanded. Many conclude the postseason is almost impossible to predict, though there are the trends to consider that are helpful. Most notably, "Small ball" seems to be a more successful approach in the postseason than the regular-season. Among teams in the postseason, the Rangers rank 3rd in stolen bases, 5th in sacrifice flies and 3rd in hit by pitch (they are however last in walks and almost last in sacrifice hits).

If you believe the Rangers will eventually regress to the mean given this disparity, it has not happened through 162 games, so statistically nothing suggests this trend will automatically change after another 19 games. In a way, the Texas Rangers have just as good a chance to win the franchise's first world championship as anybody, and that smile from Elvis Andrus will be even wider.