Fantasy football pre-season

Today I've been getting on with the essential football pre-season ritual that is choosing my fantasy football team. I've been doing this for more than ten years now, and I have to say that if there's a drab midtable dogfight on TV knowing that the end result will actually matter when it comes to end of season bragging rights makes it that bit more interesting. I also think that by making you take an interest in other teams it makes you a bit more knowledgable about the game. But anyway...

This year my knowledge is at a very low ebb. While I saw quite a lot of Match of the Day highlights last season I saw few (none?) premiership games from start to finish. This makes it hard to build up the background knowledge that makes picking a fantasy side easy. So I've been doing something I haven't done for years - dropping last seasons results into a spreadsheet and looking for clues.

This is actually a good exercise in modelling. For a start it highlights the limits of models - your only data is based on past performance, but you're trying to predict the future. The variables are often linked - but in a hard to define way.

So I'd just decided not to bother with Manchester United defenders this season as Roy Kean's retired and the midfield won't give them much protection - when I remember they've signed Michael Carrick to replace him. To work a model would have to adjust for all kinds of things.

On the other hand the model has highlighted a few things. First top players are lousy value for money. The folks who worked out the prices put a big premium on the players who score the most points - so the best investments are in the middle level players. Similarly goalkeepers and defenders offer much better value for money than midfielders and forwards.

However the goal of the game is not to produce the most cost effective team - it's to score the most points. So at some point you stop trying to save money and like Roman Abramovich on a shopping spree you realise that you're never going to win the league unless you pay waaaay over the odds for the top players. It's around about this stage that I ditched the model and started tweaking - so and so out - so and so in...

End result of all this nonsense? A starting 11 of

Van de Saar
Finnan
Kishishev
Terry (c)
Ferdinand
Carragher
Mendes
Carrick
N'Zogbia
Bellamy
Ameobi

and on the bench

Bennet
Lennon
Morrisson
Horsfield