weasel wrote: ↑Wed Mar 13, 2024 10:01 pm
For me I'd simply use my eyes and judge for myself how hard a chance it was. I'd also judge a match more by how many good chances, in my eyes, we created. That is why in Bielsa's last match, the 4-0 loss to Spurs, we weren't as bad as the scoreline suggested as I think we hit the woodwork 2 or 3 times, their keeper made some good saves and we missed a couple of absolute sitters, especially Dallas missing an open goal. By contrast Son and Kane were clinical for Spurs. We had more chances in the game than Spurs and it was decent chances too rather than 30 yard hit and hopes.
I think that's exactly why xG, or any other statistic, metric, as well as the "eye test," has to be used in conjunction with other information. The example you used is great for a leisurely weekend afternoon, or our own MOTM and player ratings, but to track large amounts of data over time the team needs something more consistent and as reliable as possible. Using our MOTM and player ratings as an example, look at how varied the results are from week to week. We'd have to have some really painful norming sessions to even attempt to get everyone on the same page--talk about herding cats
What we're also not taking into account is that undoubtedly the team have their own xG model that they adhere to. Part of the issue with xG is the various models, with various criteria, results, etc. However, if the team are sticking solely to their own model, and database of previous shots for reference, then the credibility of the statistical results is improved from mere consistency.