Expected goals or xG is one of those ‘marmite’ statistics in football, some love it and some really hate it and feel that it’s a pointless stat that’s worthless in the modern game so I tend to hesitate and limit the amount of articles we post about it.
But it’s one of those stats that is becoming more and more involved and widely used in the game. The likes of Sky Sports and BBC Sport’s Match of the Day now use it as one of their main statistics and it looks like there’s no getting away from it.
Just like any statistic when it comes to football, xG has it’s limitations – of which you’ll see in the breakdown numbers below, but it still provides some use in that it can indicate simply the quality of the chances that a team is creating. You can then compare that with actual goals scored to see how the two compare. Over-performing in expected goals (scoring more goals than expected) is ideal, it suggests that a team is taking their chances – under-performing suggests the opposite.
Not only an attacking metric, it can also be used to measure defensive performance. Are we conceding more goals than expected? Conceding less goals than expected would suggest that a team have been lucky on occasions or on some occasions facing good sides with excellent finishers.
Courtesy of infogol, we’ve broken down their xG totals for every Swansea City game this season. Simply put, every shot on goal is rated between 0 and 1 in terms of the goal expectancy of the shot.
Of our 16 matches so far, xG has been accurate in 10 of them.
| Home Team | Away Team | Score | xG Score | Actual Points | xG Points | Points Difference |
| Sheffield United | Swansea | 1-2 | 0.66-0.95 | 3 | 3 | 0 |
| Swansea | Preston | 1-0 | 1.7-0.66 | 3 | 3 | 0 |
| Birmingham | Swansea | 0-0 | 1.55-0.26 | 1 | 0 | 1 |
| Swansea | Leeds | 2-2 | 1.07-1.26 | 1 | 1 | 0 |
| Swansea | Bristol City | 0-1 | 0.48-1.86 | 0 | 0 | 0 |
| Millwall | Swansea | 1-2 | 1.28-1.09 | 3 | 1 | 2 |
| Swansea | Nott’m Forest | 0-0 | 0.28-0.86 | 1 | 1 | 0 |
| Stoke | Swansea | 1-0 | 2.16-0.7 | 0 | 0 | 0 |
| Middlesbrough | Swansea | 0-0 | 0.77-0.25 | 1 | 1 | 0 |
| Swansea | QPR | 3-0 | 1.7-0.79 | 3 | 3 | 0 |
| Wigan | Swansea | 0-0 | 1.29-1.4 | 1 | 1 | 0 |
| Swansea | Ipswich | 2-3 | 3.68-1.05 | 0 | 3 | -3 |
| Aston Villa | Swansea | 1-0 | 1.66-1.7 | 0 | 1 | -1 |
| Swansea | Blackburn | 3-1 | 0.8-1.17 | 3 | 0 | 3 |
| Swansea | Reading | 2-0 | 3.83-0.91 | 3 | 3 | 0 |
| Rotherham | Swansea | 2-1 | 1.86-1.91 | 0 | 1 | -1 |
Fixtures Breakdown
Here’s a breakdown of where xG was different from the actual match outcome, starting with the goalless draw at Birmingham City.
Birmingham City 0-0 Swansea City (xG: 1.55-0.26)
xG correctly suggests here that Birmingham City should have won this one and they definitely should have but xG points out that Birmingham’s finishing on the night was poor and our defensive got lucky to keep a clean sheet. I can imagine what some fans must be saying by this point – “football’s all about goals! You don’t win games with xG!” – and no you don’t, of course, but this is simply a indicator or a measure of the quality of chances that teams create which can then be used to determine how good or bad a team’s finishing is. Given our season so far, you would expect our overall xG for the season to be higher than our actual goals scored.
Expected result: Lose, Actual result: draw.
Millwall 1-2 Swansea City (xG: 1.28-1.09)
xG suggests here that both teams should have taken a share of the points with a 1-1 scoreline. We had ten men for most of the match but Montero came on and made all the difference with two assists in the final 15 minutes or so. Kyle Naughton’s excellent strike from the edge of the box would have scored a low xG while McBurnie’s tap in to win it would have made up the majority of our 1.09 xG score.
Expected result: Draw, Actual result: win.
Swansea City 2-3 Ipswich Town (xG: 3.68-1.05)
This is the one of the most obvious results that you would have expected here. We created a hatful of chances and xG correctly points out that our finishing against Ipswich was desperately poor. xG suggests that we should have scored at least an extra goal that we scored, while Ipswich managed three goals despite creating very little. It’s fair to say that the visitors made the very most of the little they created on goal.
Expected result: Win, Actual result: lose.
Swansea City 3-1 Blackburn Rovers (xG: 0.8-1.17)
Blackburn’s penalty in this game contributes to 0.8 of their total 1.17 xG while the Swans only managed 0.8 in total xG but were able to score 3 goals to win the game, suggesting that actually in this game, on the rare occasion they were clinical and took their chances. Fulton’s equaliser was a strike from just over 20 yards that hit the goalkeeper on its way in so that would have been a very low xG rating (0.03). Dan James slipping the ball through for Connor Roberts to score had an xG of 0.07 and Celina’s dink over the goalkeeper to guarantee the win had an xG score of 0.07. So you can see here how a team can over-perform in xG simply by being clinical and showing good quality inside the penalty area.
Expected result: Lose, Actual result: win.
Rotherham United 2-1 Swansea City (xG: 1.86-1.91)
This is the most obvious match to appear in the list of fixtures that shows a different result to what the xG suggests but still, despite our dominance, the xG suggested result is only a draw. Rotherham’s two penalties total 1.6 xG. The Swans, meanwhile, scored 1.91 xG – just short of 2 goals and they really should have netted at least twice in this game and guaranteed something before the madness at the end.
Expected result: Draw, Actual result: lose.
Overall Summary
In summary, Swansea City are not scoring enough goals that xG suggests. We’ve scored 18 goals but xG suggests that we should have scored a minimum of 24 goals – that’s 6 short. In terms of defence, we’ve conceded 13 goals and the xG suggests that we should have conceded 3 more – 16.
So in a nutshell, if the Swans can improve their finishing, they can fly up the table but that remains their main problem, as opposed to actually creating the chances – which was an issue earlier in the season.
