xG (Expected Goals)

[Dan+]

Seems to have crept into football lately, something that defines based on chances and how good the chances were, how many goals a team should've scored in a game.

Interesting theory, or load of bollocks?

[Dave]

A load of bollocks. Bournemouth and ourselves are prime examples of clubs that have scored plenty of goals in the last three seasons through an awkward bobble during a chance and having players that anticipate that bobble happening.

[Harry]

Interesting enough. Gives you a  idea where a game might have gone based on chances created. Far from perfect but a stat I'd be inclined to check out. 

[ScoRoss]

People just trying to simplify a very complicated game. Makes it easier for people to pretend they watched and understood a game.

Can't say I'm fuss about what 'advanced metrics' say.

[RandoEFC+]

Again, with the absolutely obese obsession with stats, we're slowly becoming the Americans.

I'm sure someone's put loads of work into this algorithm and I'm sure it's one of the most accurate models out there, and as a mathematician I appreciate that.

But we're talking about 22 guys trying to twat a football between two posts, sometimes the ball just bounces in off someone's arse and ruins all the carefully planned science behind things like this.

[Guest]

My judgement is through my eyes. This goes for everything in football. If a player has 20+ goals and 20+ assists but I find them to be crap, that's my opinion (Salah I'm looking at you). If a winger has 2 goals and 3 assists a season but dominates his wing creating danger and offers good support, then I'll find them to be better than the former who misses too many chances a game. Stats mean nothing, they can offer a good insight but they don't determine how good a player is.

On topic though, its all a what if situation. Its through the eyes and your personal judgement, again. If I don't jump up from my seat at a chance then I don't find it to be close to a goal, and its silly to think of the what ifs. If they get a chance and miss it then that's that. You just go on. Maybe you'll think "what would've happened if he scored this or that" but at the end of the day, if a scoreline ends 2-2 then that's it. Stats won't change the scoreline or the position of the team in the table and not only that but you'll have your own theories of the what ifs. We don't need stats to tell us that we should be thinking if this chance went in or not. 

I think Peru would have beaten Ecuador in the Copa America Centenario if Revoredo's open header went in or if Ruidiaz scored his last shot. It doesn't mean anything, it ended 2-2 and that's the result that will go down in the books.

[The Liquidator]

On 8/13/2017 at 0:14 AM, Dan said:

Seems to have crept into football lately, something that defines based on chances and how good the chances were, how many goals a team should've scored in a game.

Interesting theory, or load of bollocks?

[ScoRoss]

Heard someone on the Guardian's Football Weekly podcast talk about expected goals and how they lack context and treat everything equally. They used the example of the Tottenham vs Liverpool game earlier this season. Tottenham ran out 4-1 winners, but the expected goals were far more equal than the scoreline was.

However, how many of those 'expected goals' for Liverpool were when Tottenham were 4-1 and took the pressure off and allowed chances for Liverpool. However, they didn't score from any of those opportunities. But they all count the same. But if Liverpool score and make it 4-2, Tottenham would naturally switch back on and try and be more solid defensively as the score has tightened. But they never do as Liverpool don't score but create a good chance after a good chance.

[RandoEFC+]

1 minute ago, ScoRoss said:

Heard someone on the Guardian's Football Weekly podcast talk about expected goals and how they lack context and treat everything equally. They used the example of the Tottenham vs Liverpool game earlier this season. Tottenham ran out 4-1 winners, but the expected goals were far more equal than the scoreline was.

However, how many of those 'expected goals' for Liverpool were when Tottenham were 4-1 and took the pressure off and allowed chances for Liverpool. However, they didn't score from any of those opportunities. But they all count the same. But if Liverpool score and make it 4-2, Tottenham would naturally switch back on and try and be more solid defensively as the score has tightened. But they never do as Liverpool don't score but create a good chance after a good chance.

Same as every other statistic such as possession. In the same match I'm sure you could argue Liverpool ended up with 55% possession but mostly because Spurs eased off at 4-1 and let them have the ball.

It's why expected goals are as pointless as every other statistic besides goals, even if the measurement is arguably a more accurate version of shots and shots on target because it tries to measure how good a chance was (I.e. a 40 yard pot shot and a header saved at point blank range would both count as one shot on target but rate differently for expected goals).

[Danny]

It's a method we use so definitely not a load of bollocks. It's base day on the amount of chances you create on average and the quality behind them.

Like with most stats they're as useful as the person using them and if you don't have someone at the club who understands them properly the next the use of stats won't work.

[ScoRoss]

1 minute ago, RandoEFC said:

Same as every other statistic such as possession. In the same match I'm sure you could argue Liverpool ended up with 55% possession but mostly because Spurs eased off at 4-1 and let them have the ball.

It's why expected goals are as pointless as every other statistic besides goals, even if the measurement is arguably a more accurate version of shots and shots on target because it tries to measure how good a chance was (I.e. a 40 yard pot shot and a header saved at point blank range would both count as one shot on target but rate differently for expected goals).

Basically, people just need to just watch the game! 

[Danny]

35 minutes ago, ScoRoss said:

Heard someone on the Guardian's Football Weekly podcast talk about expected goals and how they lack context and treat everything equally. They used the example of the Tottenham vs Liverpool game earlier this season. Tottenham ran out 4-1 winners, but the expected goals were far more equal than the scoreline was.

However, how many of those 'expected goals' for Liverpool were when Tottenham were 4-1 and took the pressure off and allowed chances for Liverpool. However, they didn't score from any of those opportunities. But they all count the same. But if Liverpool score and make it 4-2, Tottenham would naturally switch back on and try and be more solid defensively as the score has tightened. But they never do as Liverpool don't score but create a good chance after a good chance.

 

31 minutes ago, RandoEFC said:

Same as every other statistic such as possession. In the same match I'm sure you could argue Liverpool ended up with 55% possession but mostly because Spurs eased off at 4-1 and let them have the ball.

It's why expected goals are as pointless as every other statistic besides goals, even if the measurement is arguably a more accurate version of shots and shots on target because it tries to measure how good a chance was (I.e. a 40 yard pot shot and a header saved at point blank range would both count as one shot on target but rate differently for expected goals).

xG will never attempt to predict every game, in fact no stat will do that because you can't. Using singular games to disprove the use of a statistic is like trying to use Hudderfield's win over United as proof as to why they're better than United. It's just not an argument.

If you're struggling to score but you're xG stats are high, then it means you're currently creating enough chances to win a game. The variable causing that could be so vast, it could be down to luck, it could be down to an influx of new players coming into the side and finding their feet, if your xG remains high as you go on game by game, then it would be fair to assume that you will push up the table.

An example of that is Brentford and Bolton, at the beginning of the season people were saying we were relegation candidates, we were bottom with Bolton. The difference between us and Bolton is that our xG stats were suggesting that we should have been winning a lot of the game we played, the stats didn't dip and eventually because of that we climbed the table and went on a 9 game unbeaten run. In fact if you go with the old adage "the only stat that matters is the scoreline" which is what a lot of people did, you'd be correct in thinking we were relegation fodder. Bolton on the other hand did not have much going their way in terms of xG and low and behold they're still bottom.

[Dan+]

I can't believe after watching those Watford highlights earlier that they only had xG of something like 1.89. They should've scored at least 5.

[Danny]

1 hour ago, Dan said:

I can't believe after watching those Watford highlights earlier that they only had xG of something like 1.89. They should've scored at least 5.

What chances are you talking about? Have they released the ratings of each chance? The goals will also be ranked with previous efforts in similar spots so what may look like an easy goal today (Will Hughes) may be missed regularly previously.

[Dan+]

7 hours ago, Danny said:

What chances are you talking about? Have they released the ratings of each chance? The goals will also be ranked with previous efforts in similar spots so what may look like an easy goal today (Will Hughes) may be missed regularly previously.

Watch the highlights of that game on MOTD and tell me they should only be scoring 2 goals from that.

[Danny]

4 hours ago, Dan said:

Watch the highlights of that game on MOTD and tell me they should only be scoring 2 goals from that.

It's the quality of the shots though, Andre Gray scuffs a shot wide when he should put it on target, I think it was Richarlson(sp?) on Watfords first counter who doesn't have get his shot on target either. Then Watford's second was a jammy own goal.

Rounding up, expecting Watford to score about 2 goals was correct from what they actually took. How many times do you think they create chances like they did this season but scuff them or don't trouble the keeper? With Andre Gray up front probably a lot.

[Inverted]

I also think xG has a flaw in that it doesn't account for situations that are brilliant opportunities but the player fails to get a shot off.

But overall it is interesting, particularly when a team seems to be playing not too bad but is struggling, and you can look at their xG to see if maybe their issues are more about luck and details than overall play.

[ScoRoss]

 

[Burning Gold]

1 hour ago, ScoRoss said:

 

It's cool to see, but I'd rather they factor in a threshold for what constitutes a 'win'. For instance, in the Fulham vs Palace game, they've given Palace the win as they had more xG (0.86-0.68), but more often than not that game would've ended in a draw.

Apparently Dom Solanke had our best xGChain/90 (the sum of xG for moves a player is involved in) and third highest xG/90 last season. Obviously sample size is an issue, but those numbers suggest he was quite unlucky last season, so it'll be interesting to see if he does any better. From the eye test, I don't see it myself, but we'll see.

[Danny]

4 minutes ago, Burning Gold said:

It's cool to see, but I'd rather they factor in a threshold for what constitutes a 'win'. For instance, in the Fulham vs Palace game, they've given Palace the win as they had more xG (0.86-0.68), but more often than not that game would've ended in a draw.

Apparently Dom Solanke had our best xGChain/90 (the sum of xG for moves a player is involved in) and third highest xG/90 last season. Obviously sample size is an issue, but those numbers suggest he was quite unlucky last season, so it'll be interesting to see if he does any better. From the eye test, I don't see it myself, but we'll see.

Sample size will be a big issue. Our owner for example says that one full season is still not enough to get the full picture of a players stats and what they mean, I think we generally look over the course of 2-3 seasons. Take into account Solanke barely played full 90 mins last season and has only been with you for a year suggests that the stats alone can't give you a detailed view on him as a player.

[Harry]

2 hours ago, ScoRoss said:

 

 wait so a goal counts for less than 1 xg? how are we less than 4...? They discounted the Mane goal for off side? How many xg did TAA's free kick count for? 

[Danny]

17 minutes ago, Harry said:

 wait so a goal counts for less than 1 xg? how are we less than 4...? They discounted the Mane goal for off side? How many xg did TAA's free kick count for? 

It's not the goal that's counted but the shot, whether it goes in or not. The quality of the shot will be given a rating and over the course of a match or season those ratings will add up which will give an indication as to how well you're doing in front of goal, the stats will give you an indication as to whether or not you should expect a drop in goals if you've been lucky, need to buy another striker even if at face value it doesn't look like you need to. They may show a that one formation or set of forwards are scoring goals but the goals aren't matching the quality of chances had, however in another formation with one or two changes the team maybe creating better quality chances that just haven't been converted.

It's just a way to analyse the quality of the chances you're creating and how often they're created vs the amount of goals you're scoring.

[Burning Gold]

17 minutes ago, Harry said:

 wait so a goal counts for less than 1 xg? how are we less than 4...? They discounted the Mane goal for off side? How many xg did TAA's free kick count for? 

Each shot is worth an amount of xG based on the probability of it being scored based on the context (position of ball, defenders, etc.). Whether it was actually scored is irrelevant to xG as it's meant to give an indication of the quality of chances created. Trent's free-kick was worth 0.05 because, although it was a good effort, a free kick from that position is a low percentage chance.

[Burning Gold]

3 minutes ago, Danny said:

It's not the goal that's counted but the shot, whether it goes in or not. The quality of the shot will be given a rating and over the course of a match or season those ratings will add up which will give an indication as to how well you're doing in front of goal, the stats will give you an indication as to whether or not you should expect a drop in goals if you've been lucky, need to buy another striker even if at face value it doesn't look like you need to. They may show a that one formation or set of forwards are scoring goals but the goals aren't matching the quality of chances had, however in another formation with one or two changes the team maybe creating better quality chances that just haven't been converted.

It's just a way to analyse the quality of the chances you're creating and how often they're created vs the amount of goals you're scoring.

I'm pretty sure the quality of the shot itself isn't considered mate. Otherwise Ruben Neves would have more than 0.14 for example.

[Danny]

Just now, Burning Gold said:

I'm pretty sure the quality of the shot itself isn't considered mate. Otherwise Ruben Neves would have more than 0.14 for example.

I meant attempt rather than how the ball is hit. Out of interest, are offside goals taken into account? Feel like Mane's second shouldn't have gained much.