Alex's Adventures in Numberland - Alex Bellos [161]
One explanation for the curse of Sports Illustrated is the psychological pressure of being on the cover. The athlete or team becomes more prominent in the public eye, held up as the one to beat. It might be true in some cases that the pressure of being a favourite is detrimental to performance. Yet most of the time the curse of Sports Illustrated is simply an illustration of regression to the mean. For someone to have earned their place on the cover of the magazine, they will usually be on top form. They might have had an exceptional season, or just won a championship or broken a record. Sporting performance is due to talent, but it is also reliant on many random factors, such as whether your opponents have the flu, whether you get a puncture, or whether the sun is in your eyes. A best-ever result is comparable to an extreme event, and regression to the mean says that after an extreme event the likelihood is of one less extreme.
Of course, there are exceptions. Some athletes are so much better than the competition that random factors have little sway on their performances. They can be unlucky and still win. Yet we tend to underestimate the contribution of randomness to sporting success. In the 1980s statisticians started to analyse scoring patterns of basketball players. They were stunned to find that it was completely random whether or not a particular player made or missed a shot. Of course, some players were better than others. Consider player A, who scores 50 percent of his shots, on average; in other words, he has an equal chance of scoring or missing. Researchers discovered that the sequence of baskets and misses made by player A appeared to be totally random. In other words, instead of shooting he might as well have flipped a coin.
Consider player B, who has a 60 percent chance of scoring and a 40 percent chance of missing. Again, the sequence of baskets was random, as if the player was flipping a coin biased 60–40 instead of actually throwing the ball. When a player makes a run of baskets pundits will eulogize him for playing well, and when he makes a run of misses he will be criticized for having an off day. Yet making or missing a basket in one shot has no effect on whether he will make or miss it on the next shot. Each shot is as random as the flip of a coin. Player B can be genuinely praised for having a 60–40 score ratio on average over many games, but praising him for any sequence of five baskets in a row is no different from praising the talent of a coin flipper who gets five consecutive heads. In both cases, they had a lucky streak. It is also possible – if not entirely probable – that player A, who is not as good overall at making baskets as player B, might have a longer run of successful shots in a match. This does not mean he is a better player. It is randomness giving A a lucky streak and B an unlucky one.
More recently, Simon Kuper and Stefan Szymanski looked at the 400 games the England football team has played since 1980. They write, in Why England Lose: ‘England’s win sequence…is indistinguishable from a random series of coin tosses. There is no predictive value in the outcome of England’s last game, or indeed in any combination of England’s games. Whatever happened in one match appears to have no bearing on what will happen in the next one. The only thing you can predict is that over the medium to long term, England will win about half its games outright.’
The ups