Everyone believes that Alf Ramsey was the most successful manager the England football team has ever had, but it simply isn’t true. Statistically, it is Fabio Capello who has been the most successful. Under his coaching, the England national team won 66.7% of all the games they played; when Alf Ramsey was in charge, they only won 60.7% of all the games they played. By contrast, Kevin Keegan only managed to win 38.9%.
Of course, it’s not how many games you win, but which ones. Alf Ramsey won the World Cup; Fabio Capello didn’t.
These are just the statistics of what happened in the past. The England football team at a 66% win rate is no better or worse than the majority of all the other top class football teams in the world. (There are exceptions – Brazil often manages to win 80% of their games, but then Brazil is somewhat exceptional.)
The problem with statistics is that they are not analytics. Knowing the average performance of the past is of little use in trying to predict the outcome of the next game. In the same way that your chances of winning the lottery remain unchanged no matter how many times you lose (and it’s amazing just how many people believe that their chances of winning increase after each time they lose). The chances of England winning their next game are 2-1, irrespective of whether they’ve lost their last three games. And the odds that their opposition will win the match are also likely to be 2-1 as well.
Analytics are different. It’s analytics rather than statistics that will help you forecast the future. Analytics and its associated modelling techniques can be used to predict future outcomes with a high degree of accuracy. Forecasting outcomes of decisions made about the team depends upon knowing which factors correlate with success and which are irrelevant and that’s the beauty of econometric modelling which, when done properly, distinguishes between causal factors and those factors which are merely coincidental.
Many businesses are increasing the use of such analytic based models to help improve efficiency and determine all the outcomes of various marketing mix scenarios. As a result they know all the consequences of any decision that they take before they take the decision, something that gives these businesses a significant competitive advantage.
For example, if the business operates in a market where launching new products has statistically only a 1 in 10 chance of success, then it is doubtful whether you would ever want to launch a new product. What analytical modelling does is identify the drivers of success for a new product, radically improving the chances of success.
Now football clubs could do the same, but of course, the football business doesn’t comply with conventional business rules. There has been some limited modelling undertaken on which factors really influence team performance and success. For example, the outlay on transfers – buying new players – explained only 16% of the teams’ variation in final league position. However, spending on team salaries explained 92% of that variation. High wages help a club much more than expensive transfers.
Unfortunately, despite the evidence, club managers want to spend ever increasing amounts of money on acquiring new players at the end of every season, a lot of whom fail to live up to expectations and never produce the outcomes that were anticipated from their purchase.
So what will be the chances of success for Roy Hodgson? Well, statistically speaking…