Expected Goals

[RumbleFox]

1 hour ago, treer said:

I suppose you could take your Xg preditions to the bookies, and use them against the odds  offered, soon enough find out if the xg predictions were any good, or a waste of time

You realise bookies use all kinds of stats to set their odds including xg? That's why they win more than not. Thanks for bringing bookies into actually, they are a perfect representation of the effectiveness of statistics. X 

[Stadt]

We’ve had more penalties than average and benefitted from a few OGs and deflections too. Defensively you could say we’ve been slightly unfortunate but our attacking approach ensures that we struggle to create consistent chances enough.

 

Is the tradeoff of barely scoring more than once a game worth it to be a reasonably competent side defensively? We’re capable of being quite good when we’re aggressive but initially we’re too cautious which is why we look better when we throw caution to the wind after conceding 

[norwichfox]

I don't have long enough life left to even try to understand xgoals, can we simply work on scored goals please? I understand those.

[treer]

22 minutes ago, RumbleFox said:

You realise bookies use all kinds of stats to set their odds including xg? That's why they win more than not. Thanks for bringing bookies into actually, they are a perfect representation of the effectiveness of statistics. X 

You are behind the times with bookmakers, no way do they use xg to set odds, they derive their odds primarily from betfair and Asian bookmakers, the day's are long gone when the UK bookmakers used many odds compilers.

A person with the skill to set and compile odds are now, either betting for themselves, or employed by the big syndicates.

As to the reason bookmakers mainly win, there are many, the biggest of which is the insentivity of punters to price offered.

, to this end bookmakers collect data, and any punter who shows signs of sensitivity to price will have their account shut,or offered maximum bets of 10pence.

[RumbleFox]

23 minutes ago, treer said:

You are behind the times with bookmakers, no way do they use xg to set odds, they derive their odds primarily from betfair and Asian bookmakers, the day's are long gone when the UK bookmakers used many odds compilers.

A person with the skill to set and compile odds are now, either betting for themselves, or employed by the big syndicates.

As to the reason bookmakers mainly win, there are many, the biggest of which is the insentivity of punters to price offered.

, to this end bookmakers collect data, and any punter who shows signs of sensitivity to price will have their account shut,or offered maximum bets of 10pence.

Thanks for the info. Some good points, well made. The only real query is exactly what your original point was? I didn't say at any point in my original posts anything to do with xg being a predictor of games, in fact I specifically said it was not a good predictive tool. But, as I said, thanks for enlightening me. X 

[Raw Dykes]

3 hours ago, treer said:

I suppose you could take your Xg preditions to the bookies, and use them against the odds  offered, soon enough find out if the xg predictions were any good, or a waste of time

No. Expected goals stats aren't very useful for telling you which team will win a game. They're not so much a useful prediction tool as a yardstick for quality of performance.

 

All they tell you is how good the chances that a team has created are. Some strikers are better than others. Some can score often while just feeding off scraps, while some can't hit a barn door in a month of Sundays, even when often being given the ball on a plate. Vardy vs. Akinbiyi.

 

I think expected goals are most useful for working out how far away a team is from success. A team may be getting poor results, but may also be creating enough quality chances to win most games. It may be the case that all is missing is a more clinical striker. Also, like Burnley last season, a team may be picking up a lot of points while creating very little. A team in this position may want to plan ahead for when the goals inevitably dry up.

 

I think of expected goals as just a more useful version of shots on goal stats. We used to just look at shots on and off target, but these can be misleading. There might be a game where one side has 14 shots, with a lot on target, and the other only has 3, all off target all game. Anyone looking at that would say that the former had the better game, but what if they had only been taking hopeful pot shots at a settled defence from outside the area and from narrow angles, easily caught by the keeper, whereas the latter broke through on goal a few times, only for their forwards to hit the woodwork or strike just wide from a few yards out. I think the latter team probably had the better game.

[Foxxed]

1 hour ago, norwichfox said:

I don't have long enough life left to even try to understand xgoals, can we simply work on scored goals please? I understand those.

I think someone on here tried to explain them to me. I've still forgotten. Something about converting good goal scoring chances.

[Raw Dykes]

Just now, Foxxed said:

I think someone on here tried to explain them to me. I've still forgotten. Something about creating goal scoring chances.

You may be overthinking it. It's simply a measure of how good the chances are that a team creates. A point blank tap in scores higher than a looping long shot from out wide, for example.

 

The factors used to work out the quality of a chance include angle from goal (much better chance of scoring from directly in front of goal than from a narrow angle), distance from goal, how many defenders there are in the way, type of shot, type of pass, etc.

[RumbleFox]

23 minutes ago, Raw Dykes said:

No. Expected goals stats aren't very useful for telling you which team will win a game. They're not so much a useful prediction tool as a yardstick for quality of performance.

 

All they tell you is how good the chances that a team has created are. Some strikers are better than others. Some can score often while just feeding off scraps, while some can't hit a barn door in a month of Sundays, even when often being given the ball on a plate. Vardy vs. Akinbiyi.

 

I think expected goals are most useful for working out how far away a team is from success. A team may be getting poor results, but may also be creating enough quality chances to win most games. It may be the case that all is missing is a more clinical striker. Also, like Burnley last season, a team may be picking up a lot of points while creating very little. A team in this position may want to plan ahead for when the goals inevitably dry up.

 

I think of expected goals as just a more useful version of shots on goal stats. We used to just look at shots on and off target, but these can be misleading. There might be a game where one side has 14 shots, with a lot on target, and the other only has 3, all off target all game. Anyone looking at that would say that the former had the better game, but what if they had only been taking hopeful pot shots at a settled defence from outside the area and from narrow angles, easily caught by the keeper, whereas the latter broke through on goal a few times, only for their forwards to hit the woodwork or strike just wide from a few yards out. I think the latter team probably had the better game.

Good post.  X

[Raw Dykes]

2 minutes ago, RumbleFox said:

Good post.  X

Thank you kindly.

[RumbleFox]

6 minutes ago, Raw Dykes said:

Thank you kindly.

You summed up what I meant in a manner I was unable to.  I'm a nice lad but not very bright.   I agree, I see xG as just a way of looking at whether a team is under or over-achieving and whether their results or performances are better or worse than one might expect and which areas (a more clinical striker) they may wish to improve on.  Of course some teams always "over-achieve" and some under in terms of xG (some teams consistently do well well very few xG and some teams consistently do poorly despite a higher xG) but, in the main, xG does seem a very good tool for assessing where a team is and whether there is room for hope of improvement or not.  As you said, in terms of predicting results it they are not the greatest tool but they are certainly worth at least casting an eye over.  I think the expected assists is quite a useful one too in terms of spotting which players are creating good goalscoring opportunities.  X

[treer]

I think we need an expected saves ratio as well,(call it Xs) because like in the example quoted( Vardy Akiynbiy), it could tell us the difference between De Gea and Pickford.

IF this came into into mainstream, we would all know who is the best keeper.

[Raw Dykes]

32 minutes ago, treer said:

I think we need an expected saves ratio as well,(call it Xs) because like in the example quoted( Vardy Akiynbiy), it could tell us the difference between De Gea and Pickford.

IF this came into into mainstream, we would all know who is the best keeper.

Yeah. I think it's the inverse of the expected goals - the higher the quality of the chance, the lower the chance of saving it. It would be interesting to see the expected saves stats for each keeper, definitely.

 

We might find that a keeper has kept a lot of clean sheets, but actually hasn't been made to work very hard for them, while others have conceded loads, but have had to deal with more than their fair share of worldies.

 

I really want to see the stats now. Good idea.

[The Doctor]

On 17/12/2018 at 23:47, treer said:

I think we need an expected saves ratio as well,(call it Xs) because like in the example quoted( Vardy Akiynbiy), it could tell us the difference between De Gea and Pickford.

IF this came into into mainstream, we would all know who is the best keeper.

The easiest way is xG against vs actual goals against. Not quite as precise (if someone spoons it over from a yard out, keeper hasn't made a save but the xG for the opposition would be pretty good because they should be scoring that). So for instance our summed xG against this season is 19.03. We've conceded 21, suggesting we're conceding goals we probably shouldn't be, but not by too much. Man City's xG against is 12.84, and they've conceded 10, so even with the miserly amount of chances they've given away, Ederson is performing higher.

[KingsX]