vorosmccracken.com

I thought an interesting experiment would be to port the National Team Rankings over to club football and see what they come up with. So as a start, I gathered up all the league and FA Cup matches for the first five divisions in the English Pyramid the last four years. Promotion and Relegation provide us with the necessary sample sizes to do this, and the FA Cup matches help increase that sample as well. I’ll start by simpling posting the rankings:

There’s lots to comment on here, but I’ll start by explaining a few things about the table. First off the rankings work the same way as the National Team rankings, however aren’t scaled the same way. IOW, you can’t take Turkey’s ranking from the National Team rankings and compare it to say Manchester City. Doesn’t work that way. These rankings are self contained. Also you may have noticed there’s a team called “Other” in the rankings. What I did is that I treated all teams from the FA Cup that did not play in the top five divisions in any of the last four years and treated them simply as a single team. It’s a shortcut that helps me keep the data while not relying on tiny sample sizes. The club’s more recent games are weighted more heavily than games three and four years ago, but you’d be surprised at how relevant results three and four years ago might still be. Finally the division listed is the division the club will be playing in this year.

Now onto the specifics. There is considerable cause for concern in using these ratings for prediction purposes. Unlike National Teams, the playing squads for club teams can fluctuate due to changes in budget and such. Wales always had Ryan Giggs, but when the budget crunch happens, teams like Portsmouth can lose all the players who compiled the results the system is using. So until I can further test how well it works, using this system as a prediction system is problematic at best. However as a system that tells you what has happened, I think it looks like it does a remarkably nice job. The clubs are nicely banded within their divisions, IE no League Two teams ahead of Premiership teams and stuff like that.

The real utility for a system like this would be for looking at the teams across Europe and handicapping which leagues are stronger than others and that will be the next thing I’ll post on this subject.

EDIT: Initially failed to credit my source for the data. A major boo-boo and I apologize.

Full credit to http://www.football-data.co.uk/ and ESPN for the source data.