SlowDownDerek asked…
We had the xG table earlier in the season that many people suggested showed we were overperforming. I think there were a few different versions. Any idea what the picture is after 46 games?
Matt Stevenson wrote…
See 2 graphs below, one which plots two lines, one for actual points and one for estimated xG points (more on this to come), and one which looks at the difference between the two numbers.
In doing this, I assumed that any result where the xG values in a game were within 0.5 of a goal of each other could be called a draw, otherwise, the team will the higher value would win. It’s completely arbitrary but felt reasonable.
Looking at the difference graph, you can see that in the early part of the season, we had been riding our luck, and after game 15, (Reading away) we had 11 more points than xG suggested. However, we then had a run of results that ‘should’ have been better, and after game 33 (Peterborough at home) xG estimated points were only 1 lower than actual points.
In the end, we finished with 5 points more than ‘expected’, which is probably a combination of luck, good goalkeeping and clinical finishes. So, I don’t think there is any compelling reason to say we didn’t deserve to finish second, although after 15 games there were concerns.
Coobster wrote…
I find the overall xGF and xGA much more interesting than the points. As Matt mentioned in his, what figure do you use for a draw turning into a victory\defeat.
So, West Brom forgot how to score most of the season? That is quite a lot of missed goals! (Unless I have misread the table!)
SlowDownDerek replied…
There must be a pattern or reason for it. Loads of low-quality longshots boosting the numbers maybe? They did have more shots and you would probably say we were guilty of trying to find the perfect chance more than we might have done. I suppose the proof is in the pudding.
Coobster added…
I was thinking the same thing about long shots. In theory, it shouldn’t matter if you have 100 1% chances vs 10 10% you should get 1 goal.
The law of large numbers only applies once you get 10’s of thousands I believe, so many poor percent shots may not be the best way of going about things when you get around 600 shots a season (looking at the bein sports stats). Obviously, you could get super lucky (or skilful?) and end up scoring loads…
With stats and the like being used more and more in football, I wonder what conclusions managers are drawing from this. It felt like under Eddie in the Prem, especially the latter years and Parker this year that we didn’t take too many long shots, but seemed to do more last season. Not that I have any proof of this. – Join the conversation, click here.
