Trying to predict team and player performance is part of the fun in FPL.
In this article I take a look at expected goals from recent seasons to see what it reveals about team and player performance.
Expected Goals – Yikes!
Oft-quoted and sometimes divisive, expected goals lacks the “Ronseal” transparency of other statistics: “Shots on target” and “Touches in the penalty area” do exactly what they say on the tin.
Expected goals is a little more abstract, it’s defined as a measure of chance quality. It aims to quantify the likelihood of a goal being scored from a particular shot, where 1 = a certain goal and 0 = a certain miss.
For instance, if Kane had eight shots in a match xG tells us how good those shots were. In this sense expected goals is descriptive. It helps paint a picture of a football match.
However expected goals can cause confusion and derision, especially when it is completely at odds with the scoreline:
Part of the problem is the name. Perhaps if “expected goals” was called “chance quality” then it would be easier to grasp what the metric is telling us.
The reasons why a team overperforms or underperforms their expected goals is open to interpretation. As Bobby Gardiner observed:
The beauty of football is often in its randomness — the team who has the better chances will in theory win the match more often, but not always.
Expected goals is also predictive. Or rather, expected goals has been proven more predictive on future team performance than any other metric. Not by much – total shot ratio is pretty good too – but consistently and by enough to be significant. Mike Goodman explains:
[xG] was designed to take lots of information, thousands and thousands of shots, synthesize it, and use that information to represent how many goals a team might reasonably be expected to score or concede given the types of shots they’ve taken and given up.
So while a single match can be fairly random, over the long term teams end up roughly where you’d expect them to be.
4 years, 4 months ago
Superb, superb article Top Marx. Brings together a lot of key issues from this IB and sets some of those common xG misconceptions to bed!
If anyone wants to post xG data in reply to this article, post it and then flag it to the team as members only via the report function.