This is from The New York Times, Science Times section, July 8 2014.
By John Tierney
I’ve been watching the World Cup with some frustrated American social scientists. When they see an underdog team triumph with a miraculous rebound or an undeserved penalty kick, they don’t jump up and scream “Goooaaalll!” They just shake their heads and mutter, “Measurement error.”
If you regard a soccer match as an experiment to determine which team is better, then it’s not much of an experiment. It involves hundreds of skillful moves and stratagems, yet each team averages only a dozen shots, and the outcome is decided by several quick and often random events. In most games, no more than three goals are scored, and the typical margin of victory is a single goal.
To a scientist, the measurements are too few to draw a statistically reliable conclusion about which team is more skilled. The score may instead be the result of measurement error, a.k.a. luck.
That can make soccer seem terribly unfair, at least to many Americans accustomed to higher-scoring sports. We don’t understand why the rest of the world isn’t clamoring for a wider goal or looser offside rules or something to encourage more scoring.
But if the rest of the world took our helpful advice, would soccer really be any fairer? Not necessarily, say the economists and statisticians who have been analyzing the balance between skill and luck in sports and in the rest of life.
Because of fluke goals, low scores and the many matches that end in ties, soccer is less predictable than other major sports, as Chris Anderson and David Sally explain in their soccer book, “The Numbers Game.”
The authors, who are professors at Cornell and Dartmouth, as well as consultants to soccer teams, found that the team favored by bettors won just half the time in soccer, whereas the favorite won three-fifths of the time in baseball and two-thirds of the time in football and in basketball. After surveying the research literature, they concluded that a soccer match’s outcome was about half skill and half luck.
But just because an individual soccer game can be decided by a lucky bounce doesn’t mean that the game is less fair than other sports.
There’s another factor to consider: the paradox of skill, as it’s termed by Michael Mauboussin, an investment strategist and professor at Columbia, in “The Success Equation.”
Suppose the world’s best Scrabble player, which would be a computer, competes against a novice. The computer’s skill will routinely ensure victory even if the novice draws better tiles. But if that computer plays an equally skilled opponent, an identical computer running the same program, then the outcome will be determined entirely by the luck of the draw.
That’s the paradox of skill in sports, business and most other competitions: As the overall level of skill rises and becomes more uniform, luck becomes more important. Mr. Mauboussin has calculated that luck matters less in English soccer’s Premier League than in the N.F.L. and in Major League Baseball, because the American leagues have evened the level of skill among teams by sharing revenue, imposing salary caps and giving better draft choices to the weaker teams.
Soccer in the rest of the world doesn’t have these constraints, so there are much bigger disparities in teams’ skills. In league play, rich clubs like Manchester United, Real Madrid and Bayern Munich buy the best talent. In the World Cup, the larger, more affluent countries can lure the best coaches and draw from a bigger pool of talent.
“Of all the major team sports, soccer is the most unequal in the sense that teams with vastly different resources regularly compete against each other at the highest level,” says Stefan Szymanski, an economist at the University of Michigan and a co-author of “Soccernomics.” If matches were purely contests of skill, the many David-and-Goliath games in soccer would be boring — and seem unfair in another way.
“If you doubled the size of the goal, then soccer would become like basketball, and in a high-scoring game, the rich teams would almost always win,” Dr. Szymanski says. “Randomness favors the underdog. Would we ever want to reduce the role of luck in soccer? No way.”
Still, some forms of soccer luck just seem dumb, like the flip of a coin before a penalty shootout that determines which team goes first in each round. The first kicker makes the shot about three-quarters of the time, which puts pressure on the other team’s kicker to even the score.
That added pressure is presumably why the team going second wins the shootout only 39 percent of the time, according to Ignacio Palacios-Huerta, a game theorist at the London School of Economics and the author of “Beautiful Game Theory: How Soccer Can Help Economics.”
In experiments with professional soccer players, Dr. Palacios-Huerta found that the odds became more even in a penalty shootout if one team led off in the first and fourth rounds, and its opponent led off in the other three. Dr. Szymanski prefers a different shootout modification to further help the underdog: Let the lower-seeded team go first in all the rounds. Either change sounds like a good idea.
So does an innovation from American sports: peer review. It could reduce the most maddening form of soccer luck, which occurs when a penalty kick is wrongly awarded after a player pretends to be fouled near the goal. No other sport gives players such an incentive to scam the referee. Before awarding a penalty kick that may well decide the match, officials could at least review video replays to make sure the referee saw a foul instead of a flop.
Over the long haul, with enough measurements over the duration of a season or a World Cup, skill does prevail in soccer. The law of large numbers limits the underdogs’ lucky streaks. League championships and the World Cup are repeatedly won by the same few powerhouses, because it takes skill to endure.
But the outcome of any one match is unpredictable enough to confound the most sophisticated computer modelers, as Roger Pielke Jr. of the University of Colorado has demonstrated in his evaluation of a dozen forecasts for this World Cup. He found that the “stochastic model” of Goldman Sachs economists and the elaborate Soccer Power Index developed by Nate Silver of FiveThirtyEight made fewer correct predictions for games in the group stage than did much simpler systems based only on the monetary value of the players or on the teams’ ranking by FIFA, soccer’s world governing body.
The best forecasters turned out to be a team at Danske Bank in Copenhagen and a software engineer named Andrew Yuan. But they were still wrong about 16 of the 48 games, and they identified only 11 of the 16 teams to advance past the group stage.
No matter how much number crunching the quants do, no matter how skilled a team is, there’s just no way to anticipate the measurement errors in each match. The forecasters, like the players, may complain about their bad luck, but it’s a fortunate state of affairs for the fans, especially those who root for underdogs like the United States.
