Soccer Outcomes Up In The Air Statistical physicists at LANL find the underdog is more likely to win in increasingly popular sport JOHN FLECK Journal Staff Writer 5 January 2006 Albuquerque Journal Soccer fans, take heart: Your sport is more interesting than football, basketball, baseball or hockey. This is no longer some unresolvable bar bet. A team of Los Alamos physicists has proven that, by one measure at least, British "football" has historically been the most interesting professional sport. Baseball came in a very close second in the study by Eli Ben- Naim and his colleagues. But if you want to win the bar bet, everyone has to accept Ben- Naim's definition of "interesting" -- a sport where the underdog has the best chance of upsetting a more dominant team. That is what happened last November when middling Doncaster prevailed over Manchester City, then Aston Villa on the way to Britain's coveted Carling Cup. "Amazing style," enthused Australian sports writer James Willoughby of the upstart's victory. In over a century of games, the underdog -- the team with the worse record going into a game -- won 45 percent of the time in British football, Ben-Naim and his colleagues found. Baseball was a close second, while National Football League and National Basketball Association games were the worst, at 36 percent. Lately, however, baseball and American football have been getting more competitive, with more underdogs winning, while British soccer has been drifting in the opposite direction, Ben-Naim noted. The research was not about winning bar bets. It is part of an attempt by Ben-Naim and his colleagues -- Argentinian graduate student Federico Vazquez and Boston University professor Sydney Redner -- to apply the tools of statistical physics to some very human questions. The question, Redner recalled, first came in a serendipitous conversation years ago with a sociologist: "How do some people climb up to the top of the social ladder and other people end up on skid row?" That might seem an odd question for a statistical physicist, normally more interested in things like the properties of molecules in a gas. But once armed with the statistician's tool kit, Redner said, intriguing questions are all around. In the same way that there are faster- and slower-moving molecules in a gas, there are winners and losers in social and economic competition -- the rich and the denizens of skid row. Redner, who worked with Ben-Naim and Vazquez on the problem while on a fellowship at Los Alamos last year, said the question of winners and losers in human interactions is tailormade for the statistical physicists. Ben-Naim, Vazquez and Redner took a first crack at the problem by building an idealized world of human interactions and turning them loose in a virtual world to see what would happen. Their virtual world ended up with things like a static underclass and an upwardly mobile middle class, just like in the real world. The problem with the model, though, was that it was a complete abstraction. How, the scientists wondered, could they find some real- world data to match their models against? The answer, Redner realized, was in sports, where obsessive record-keeping has created a wealth of data on human competitions. The scientists used British Football Association records back to 1888, and 60 to 100 years of data for the other sports. "Sports is a place where the data is very accurate," Ben-Naim said, "and there is a lot of it." Sports is not exactly real life, but it matches key elements of real life. "In societies, people compete," Ben-Naim said. And then there's the question of the sports themselves, which is not lost on Ben-Naim, who described himself as a soccer fan. "This is one way to rate sports," Ben-Naim said. "You can think of it a little bit as a 'Consumer Reports' for sports."