Steven
Active Member
From the Times.
Weathering the storm that is predicting the Championship
Chris Iwelumo of Charlton battles with Patrick Kisnorbo
Daniel Finkelstein
There is something important to understand about making predictions. The first step towards making a good one is to realise that predicting things is not easy.
Take the weather forecast. You may shout at the news, but I shout at the weatherman. As they cheerfully say that it will be sunny tomorrow, I yell: “How do you know that? Are you God? And if you are, can’t you get a better job than presenting the weather, for heaven’s sake?”
What the presentation of Met Office data should say is how likely it is to be sunny tomorrow. This is more useful and more truthful.
And this is how the Fink Tank works. The predictions we make are simply probabilities that accept the idea of randomness. Or, to use a more conventional terrace term, luck.
Where these probabilities are an improvement over guesswork is largely that the computer model can integrate all the possible outcomes in a way of which the human mind is not capable. Even Stephen Hawking or Robbie Savage would struggle.
Take the Coca-Cola Championship. Let’s say that sixth-placed Burnley have a run of good luck (and their season has got off to a good start, with six points more than we thought they would have by this stage). It may be possible to calculate in your head the rough impact on them. But what about the way it affects all the teams around them?
So Dr Henry Stott and Dr Ian Graham, using goals and shots on goal from the past two seasons, use computer modelling to take account of lots of different pieces of good fortune. By simulating the league season many tens of thousands of times, you can see, for instance, how often Burnley win automatic promotion. And on the graphic you can read the answer - 5.2 per cent of the simulations ended with Burnley in one of the top two slots. Or to put it another way, they have a 5.2 per cent chance of going up without a play-offs match.
The most striking feature of the work is how strong West Bromwich Albion are. We give them an extraordinary 71 per cent chance of automatic promotion. Almost as surprisingly, Watford, despite being top, have only a 24.4 per cent chance of finishing first or second. Charlton Athletic’s chances are 44.9 per cent.
As for the rest, there is an extraordinary spread of teams with a good chance. Half the teams in the division have a 20 per cent or greater chance of ending up in the play-offs. The competition is thus much tighter than the Barclays Premier League, where the class of the teams fall off sharply after the big four are taken into account. The best team in the Championship (West Bromwich) are a third as good as the best team in the Premier League (Manchester United) and three times as good as the worst side in the Championship (Barnsley). Meanwhile, United are more than six times better than Derby County, the bottom club.
A warning not to get overexcited about a few results is provided by Coventry City. They have scored 5.6 points more than we expected them to have at this point. And then they won at Old Trafford in the Carling Cup. Yet guess Coventry’s chances of getting in the top six. Only 4.9 per cent.
One thing to take into account is that luck bounces about. There have been terrible starts for Southampton, Preston North End and Queens Park Rangers, while Charlton and Watford have had terrific starts. This does not mean that their luck is bound to change, merely that, with a close league like the Championship, it can all look very different, very quickly.