HERE IS A personality test. In 1963 Don Revie asked a gypsy to come to Elland Road to exorcise Leeds United’s ground. Do you believe: (a) that this successfully lifted an ancient curse and led to more than a decade of glory for the club or (b) gypsy schmypsy. The definitive book on the Revie era is called The Unforgiven; to judge from this story it should be called The Unhinged. The glory years came from talent and gamesmanship.
Now, if you agreed with the former you are a prize chump only fit to be manager of England. If you opted for the latter, then well done, you are very perceptive.
Before you come over all smug, however, you might ask yourself whether there are some superstitious footballing myths that even you accept. Do you believe: (a) that teams have winning streaks that make them more likely to triumph on their next outing or (b) winning streaks make no difference to the probability of victory next time? If you answered (a) you hold a position which almost everyone in the sporting world shares and which pops up again and again in analysis of games. If you answered (b) you would be right. This means that you are either an academic statistician or someone who cheated because you realised that I wouldn’t have bothered asking the question unless the answer was (b).
There is a copious and fascinating academic literature on the subject of winning and losing streaks. The classic study by Gilovich, Vallone and Tversky was published in 1985 and analysed a season of an American basketball team’s shooting.
The authors pointed out that if, for instance, a particular player had a 50 per cent chance of scoring from a shot, and if scoring with one shot made no difference to his chance with the next shot, you still wouldn’t expect that success to be evenly spaced — miss, hit, miss, hit and so on. You would get streaks of hits and misses. The question is whether, in practice, there are more and longer streaks than you would expect to find if each of the shots were truly independent of the next.
Guess what. There weren’t. The streaks of form made no difference.
Analysis of football produces the same conclusion. Alex Morton and Henry Stott’s work for The Times Predictor shows that adjusting for form over the past few games does not help the model. When Peter Ayton, of City University, analysed the records of the top 12 Premiership goalscorers in 1994-95 and 1995-96 he found that if a player had scored in his past few games it gave him no more or less chance of scoring in his next one.
Being given the manager-of-the-month award is famously regarded by fans as a disaster, because all too often the team promptly begins to lose. The data provides a possible explanation. The award is given after a cluster of games in which a side does better than expected. This may have occurred without any change in the team’s ability to win. It isn’t all that surprising if it is followed by a return to poorer results.
Looking at 70,000 games has even led one group of economists to argue that both good spells and bad ones are shorter than should be expected. They suggest that the reason for this is that teams with a few wins under their belts become complacent, while losing teams try harder.
Some other research provides an alternative way of looking at it. Academics following ten-pin bowling and championship horse-shoe pitching have detected the real runs of form missing in basketball and football. Perhaps football players and teams on a roll have their form disrupted by more determined opponents. This is, of course, impossible in ten-pin bowling, accounting for the different research finding.
During a batting slump, David Gower wore a T-shirt with the words “form is temporary, class is permanent” emblazoned on it. He was overstating the case for form. Class is everything. Unless, that is, you pitch horseshoes for a living.
