Post Match Leicester 2 Tottenham 3

[fitz]

You're comparing apples and pears. Unless Lawro starts watching matches and, after the game, analyses the statistical likelihood of every chance being scored/conceded, and therefore calculates an 'expected' result, I don't see how that's relevant.

One is a forecast, and one is a historical data set.

And for the record, feel free to calculate how many points away Lawro has been for us for the last six seasons on average. I'm happy to bet 1,000 Joe_Fox bucks that it's more than 0.3pts per season on average.

Aye. But they go back and analyse the chances. Then say that every game the chances Brighton create should lead to xg4 yet they score 0 or 1. At some point expected misses have to come in to the equation? Unless Brighton are now on 90+ chances a game.

 

[popillius]

Aye. But they go back and analyse the chances. Then say that every game the chances Brighton create should lead to xg4 yet they score 0 or 1. At some point expected misses have to come in to the equation? Unless Brighton are now on 90+ chances a game.

Honestly. It's sometimes painful trying to explain this shit.

 

[popillius]

You're comparing apples and pears. Unless Lawro starts watching matches and, after the game, analyses the statistical likelihood of every chance being scored/conceded, and therefore calculates an 'expected' result, I don't see how that's relevant.

One is a forecast, and one is a historical data set.

And for the record, feel free to calculate how many points away Lawro has been for us for the last six seasons on average. I'm happy to bet 1,000 Joe_Fox bucks that it's more than 0.3pts per season on average.

Or, I'm suggesting that a forecast by 1 person based on nothing more than his whim is, season by season. About as accurate as xG.

 

[bocadillo]

Honestly. It's sometimes painful trying to explain this shit.

It's painful reading it as well.

 

[Blue Maniac]

Where can I spend Joe_Fox bucks? Just at the circus?

 

[FryattFox]

Or, I'm suggesting that a forecast by 1 person based on nothing more than his whim is, season by season. About as accurate as xG.

But xG itself isn't trying to forecast anything.

It can be used to assess over- or under-performance to a statistical mean. In which case it is relatively accurate over a large dataset (e.g. a much larger dataset than just 38 matches). There will always be anomalies, some of which can be explained (it's pretty clear to anyone that watches Brighton matches that they lack even an average striker).

If you're actually comparing Lawro's predictions to a model based on a database of over 100,000 shots, you must have a high opinion of Lawro.

 

[popillius]

But xG itself isn't trying to forecast anything.

It can be used to assess over- or under-performance to a statistical mean. In which case it is relatively accurate over a large dataset (e.g. a much larger dataset than just 38 matches). There will always be anomalies, some of which can be explained (it's pretty clear to anyone that watches Brighton matches that they lack even an average striker).

If you're actually comparing Lawro's predictions to a model based on a database of over 100,000 shots, you must have a high opinion of Lawro.

No. I have a very low opinion of xG.