BLUEBIRDS AND SWANS... LATEST NEWS & VIEWS
NOT many complained with Paul Clement’s scathing assessment of his own Swansea City side at the weekend.
“I believe we’ve gone bottom of the table today and I can’t say that we’re unlucky,” said Clement following the defeat at Stoke that saw the Swans hit the foot of the Premier League, accurately describing the team as “toothless” in the process.
“We’ve lost 10 games now, it’s not been through misfortune: ‘Oh, we’re the unluckiest team in the world’.
“We deserve to be bottom because we’re losing too many games: we’re not consistent enough through a 90-minute period.”
The league table, after all, doesn’t lie.
Maybe so, though there are those who are at least holding that old adage up to statistical scrutiny.
Expected goals (or xG for short) has become increasingly turned to over the last few years as a tool to more accurately analyse a team’s performances instead of looking simply at shots on goal or greater possession.
But it’s received a boom this season having been brought into the mainstream by the BBC’s Match Of The Day.
There are different models, but in essence it is a way of measuring the quality of a chance created in a match. Less Expected Goals perhaps, and more Probable Goals.
Several factors are taken into account to come up with the ‘quality’ of the chance, such as distance from goal, the angle of the shot and so on. A one-on-one in a central position in the area, for example, will have a higher xG value than a shot from outside the area.
The better the chance, the higher the number – and the more of them, the greater the xG value for a side in a match.
It does help to look past results to analyse performances to see whether teams are getting ‘lucky’ or not. A side with a low xG in a game but more goals actually scored could be down to a defensive error or a bit of individual magic. A side with a high xG value in a match but with less goals can be the victim of a goalkeeper in fine form or some poor finishing.
In other words, it asks questions of whether good form is sustainable or not – and whether bad form is down to bad luck or just bad play.
For example, early advocates of xG highlighted Swansea’s record-high finish of eighth under Garry Monk in 2015 as over-performing according to the numbers, getting away with wins when the statistics didn’t back that up. The struggles the following season proved it right.
The struggles haven’t stopped since.
The xG for last season suggested Swansea should have not escaped the bottom three – perhaps indicating the individual influence of Gylfi Sigurdsson in their survival. But what of this season? With 15 games gone, there are few thinking that Swansea have been robbed of results. Indeed, there might be those who would argue that they are fortunate to have nine points.
But then there have been a few lastminute winners for opponents that may have skewed some of the stats. really
Both cases are supported by the Expected Goals model used by Gracenote Sports that suggests Swansea ‘should’ have four more points than they do – and be sat 16th in the table.
“Our model is basically similar to others in that we estimate the chance quality based on a variety of variables including distance, angle, header or shot and game situation (buildup, set-piece, counter etc),” says Simon Gleave, head of analysis at Gracenote Sports.
“The difference to others is how we compile the table in that we allocate points – three, one or zero – based on the xG difference between the two teams in a match.
“Looking through the matches, Swansea basically have four points fewer than xG says they could have.”
It’s not much to shout about. Indeed, there are those aforementioned games where Swansea were fortunate to come away with points.
The xG model suggests they could have lost by two to Southampton on the opening day instead of holding out for a goalless draw, just as the statistics suggested another day would have seen a win for Tottenham at Wembley. Likewise, there were enough quality chances for Crystal Palace to have earned a draw in Swansea’s August win at Selhurst Park.
Conversely, the model does offer Swansea – and Clement – some sympathy. Defeats to Newcastle and Burnley and Brighton could all have been draws, while wins could – or should – have come against Watford and Bournemouth. The model would argue, for example, that Leroy Fer would normally score the chance that would have earned victory against the Cherries.
Across all the clubs, Swansea are one of five to be more than three points under the tally xG suggests they could have, with Arsenal, Spurs, Southampton and Crystal Palace the others. Indeed, having lost out in games where they have had a flurry of chances, the confidence of a Palace revival is backed up by the statistics that says they could have 10 more points.
While the statistics say Swansea may be a little unlucky, that is as good as it gets. Ultimately, the chance creation is still far too low for anyone to start clutching at straws (more than two chances per game down from last year) and there is noone able to buck the trends by providing bits of individual flair.
The hope it does offer is that some others – Huddersfield a prime example – may not be able to keep up fighting against the xG averages as they have so far.
But the only statistic that matters for Swansea – and Clement – right now is the one on the scoreboard.