Probability
Imagine a team that has played 38 games, a full season. They won 13, drew 13, and lost 12. Fairly average. Not out of the question for Fulham, although perhaps next year.
How did that team get those results?
Here are five random sequences.
DDDLDWLLWDDL DWWWWWLWLDLDDLDWDWLWLDLLWW LWLLLLWLDLDD DWWLWLDDDDWWWWDDWLWLWDWLDD WLDWDLDWDWDL DLWLWWLWLDWLDWWDLLLDDDWWDL DWWDDWDDWDDD LWLDWWDWWDWLLLWLLLLLDDWLWL DLDDDWDWLWWW DLWLLWLDWLWWDLWWLLWLDDLDDD
All five teams are exactly the same. They end with the same number of points. If you were to judge each team after 12 games you would come to vastly different opinions.
Team 1 looks a bit like us this year.
Team 2 looks like it’s in a bit of trouble.
Team 3 is doing okay.
Team 4 hasn’t lost yet.
Team 5 is doing quite well too.
This is stretching a point – football isn’t random – but equally, it does show how one team could get to the same point in very different ways, and that, by judging the team after 12 games, we’re missing the bigger picture.



Very true! Unfortunately, this is the central problem with stats in sport – one is always working with a sample size that just isn’t big enough!