The Benefits and Downfalls of Sports Analytics

Photo courtesy of Taylor Gray, Student Publications

I love reading books written by journalists, and I especially love it when those books revolve around sports. When I picked up Ben Reiter’s “Astroball”, published a little under a year ago, I knew the book would provide at least a few hours’ worth of entertainment. Reiter is employed by Sports Illustrated and the book chronicles his journey following Major League Baseball’s Houston Astros. Over the course of a few years, the Astros go from lovable losers to world champions. In many ways, the story would ring familiar to anyone who has read or watched “Moneyball” – underdog team finds castaway players with under acknowledged talent and morphs the group into a winning strategy. The analytics department plays a key role in this transformation. The story expresses that baseball will
never be the same.

Yet there is something notably different about the way Reiter – and the front office personnel he interviews – describes the team’s turnaround. In the world of “Moneyball”, there is no room for old-school baseball scouts. As movies are wont to do, the film resorts to caricature. Oakland’s longtime scouts are hopelessly out of touch. They do not know who Fabio is. The scouts dock a player of points for having an unattractive girlfriend. They focus on form over function. Why, general manager Billy Beane wonders, won’t these men focus on what matters: getting on base?

If Reiter’s book is not a direct rebuke of the analytics-or-else mentality, it is at least a strong reminder that not every valuable trait can be boiled down to a statistic. Astros general manager Jeff Luhnow is a character who would have fit into the “Moneyball” script neatly. An Ivy League undergraduate turned management consultant turned entrepreneur, a reader expects Luhnow to take the Beane role – that is, to relentlessly shepherd the Astros organization towards big data and advanced metrics, over the protestations of behind-the-times baseball veterans. Instead, Luhnow and his team work tirelessly to meld analytics with the traditional scouting approach. The approach does not always work. In one case, the Astros cut a player who tried to convince them he had made significant improvements over the offseason. His career numbers, they decided, did not suggest that he could amount to much as a major leaguer. Quickly picked up by another team, he immediately proved himself. The Astros lost a budding young superstar for nothing.

But in some cases, the approach paid off in dividends. The Astros signed a declining Carlos Beltran, a grizzled veteran who was, by virtually every metric, nearly out to the pasture. He was the sort of player “Moneyball” trains fans to give a wide berth. Perhaps he was once a good player, the statisticians say, but to sign him to any meaningful contract now is to pay for past performance. By signing Beltran, the Astros found a key leader for a young team, a player who knew what it was like to win – and lose – a playoff series. They found in Beltran a coach who could spot and correct mistakes his teammates were making in real time. And above all, they found a stabilizing force in the locker room. No existing statistic could account for the value that did not appear in statistical

As I read “Astroball”, I was reminded of how unlikely this story seemed to nearly everyone in 2014. That June, Reiter made the front cover of Sports Illustrated with an article declaring the Astros 2017 World Series champion. To many – especially Astros fans – it felt like a cruel joke. How could a team mired in such hopelessness reach the top so soon? It urged me to consider the misconceptions we might carry with ourselves in everyday life. One thought that is particularly applicable at Tech is the belief that some combination of the hard sciences and engineering can solve every single problem. World hunger is a genetically modified crop away from extinction. The skeleton of a perfect dating site is hidden on an undergraduate’s hard drive. But, what good is all the food in the world if we cannot distribute it? Isn’t a social network only as strong as the people who use it? At the end of the day, our problems are human ones – in sports, in politics and beyond. Reiter shows us that regression models and probability tests can only take us so far. The rest must be solved through processes that are as imperfect as we are.