Premier League table – by expected goals: why everything you thought about this season is wrong
The table lies. Obviously the literal points tally is true enough, but, contrary to popular belief, the table does lie – about how good or bad each team really is. Also, Santa isn’t real. Sorry.
xG, as you probably know by now, refers to expected goals. It’s a metric that analyses the quality of shots taken by each side in a match and, using tens of thousands of simulations, calculates a probable scoreline on the basis of those goalscoring chances. Here, read this if you like.
As xG has become better-known, to the unfettered rage of Craig Burley, Jeff Stelling and your dad (assuming neither of those is your dad), its use has spread. Using xG and xGA, which analyses the scoring chances a team allows their opponents, it’s even possible to work out "expected points", or xPts. At FFT we quite like Understat, who have created a league table to reflect what we’ve all seen on the pitch, and not just the goal tallies at full-time.
Here’s a fictitious example of how it can work. Arsenal beat Manchester City – hey, it could happen – by a 1-0 scoreline, though City dominated the game and xG data suggests, based on its many simulations, based on the quality of shots taken, that a more probable scoreline would’ve been Arsenal 0.39-2.11 Manchester City (we’re stretching the limits of the word ‘probable’ here). Play the same game many thousands of times and City would have won 86% of the time, Arsenal just 3%, with 11% drawn. Arsenal have the all-important three points, but only 0.2 xPts. Why? Because they took three points in 3% of simulations and one point in 11% of them, and (3 x 0.03) + (1 x 0.11) = 0.2. Manchester City had 2.68 xPts but still lost, because it sucks to be them.
Caveat time: obviously xPts isn’t perfect, and it is less precise than xG. However, it is broadly indicative of patterns we see with our own eyes, and even cynics may begrudgingly acknowledge the general truth of what’s below, unless they think the current Premier League table perfectly reflects the balance of play in every game.
And, helpfully, we start with a perfect example of how the league table doesn’t tell the full story…
20. Newcastle (real position: 12th)
If Steve Bruce could consider himself lucky to be given another Premier League job – not that he would – it’s nothing compared to the good fortune he’s had since. Nobody would be surprised to see this Newcastle side in the bottom three and yet there they are in mid-table, needing only a couple more wins to secure safety.
We can’t rule out the possibility that Bruce is a witchdoctor. Newcastle have gained 14 more points than might be expected on the balance of their matches, despite posting abysmal shot data all season. No team has had fewer shots from inside the box. Only one takes a higher percentage of their shots from outside the box. And yet, just as half of their starting XI’s outfielders are defenders, exactly half of Newcastle’s goals (12/24) have been scored by defenders.
Premier League Shots in opposition box per game vs % of Shots in box taken (shots in box/total shots) pic.twitter.com/3jKFBz3eCf
— BA Analytics (@Blades_analytic) February 4, 2020
And not only do Newcastle have the Premier League’s worst xG, they also boast the third-worst xGA. Fortunately for them, Martin Dubravka has been in inspired form: Newcastle have conceded 10 fewer goals than could be considered probable, based on the quality of chances presented. It’s the biggest difference in the division.
19. West Ham (18th)
That’s right. West Ham are even worse than they look.
18. Aston Villa (17th)
While it was a huge blow to lose Wesley for the rest of the campaign, a similarly season-ending injury to Tom Heaton could be what sends Aston Villa down. They’ve allowed the highest xGA of any club; Heaton was preventing a bad situation from becoming even worse.
17. Norwich (20th)
Promoted sides generally need fortune to favour them. Alas, Lady Luck has kicked the Canaries down a mineshaft scheduled for demolition. Although it’s true that every team must contend with injuries, Norwich had more than one player ruled out for most or all of a season that had barely begun. At one point they were without four of their five centre-backs, and when they beat Manchester City it was with eight players missing.
Of course, Daniel Farke’s charges would be candidates for the drop anyway. They concede goals too easily, and only two players have scored more than once in the league. Still, it’s a little harsh that relegation already seems inevitable, with Norwich facing top-eight sides in each of their next four fixtures, before trips to Arsenal, Chelsea and a wounded Manchester City. Gulp.
16. Crystal Palace (14th)
Just 0.04 xPts ahead of Norwich (stop rolling your eyes) are Crystal Palace, who had gone into the October international break in sixth place, having played eight, scored eight and conceded eight. Clearly that was never going to last, but it’s been an alarming slide since then, in the xPts table and actual points table alike.
The immediately obvious problem is goals, or a lack thereof. Palace haven’t scored more than twice in any game this season, whatever the competition, and relying on fine margins can take you only so far. Their league-low 22-goal tally is down to a shortage of chances being created, rather than good ones being missed.
However, as their 12-point lead over Norwich shows, things could be even worse. Vicente Guaita has been quietly excellent in 2019/20, helping Palace to concede nine fewer goals than their xGA, but as last weekend showed, they can’t rely on him forever.
15. Bournemouth (16th)
Unlucky with injuries; not so unlucky in games. And some would argue that Bournemouth’s injury problems aren’t just bad luck anyway. At least Aaron Ramsdale has settled quickly into Premier League life, letting in fewer goals than he ‘should’ based on the quality of chances.
14. Burnley (11th)
One (real) point fewer, yet three (real) places better off – it’s a tight league, all right.
Although Nick Pope hasn’t been as magnificent between the sticks as he was in 2017/18, when he helped Burnley to concede one-third fewer goals than their xGA and ultimately qualify for the Europa League, he’s still an asset.
13. Watford (19th)
Blimey. A club that looked utterly doomed at the start of December, dead last with 8pts from 15 games and appointing their third manager of the campaign, have been transformed by said manager. Nigel Pearson lost to Liverpool in his first game, then took 13 points from five matches. Watford aren’t out of the woods yet – clearly, seeing as they’re 19th – but, like Pearson facing a pack of wild dogs, they look capable of surviving unlikely odds.
What’s more, the xPts table suggests Watford, in the exact words of Beverley Knight, shoulda woulda coulda had 33pts right now. Instead, they have 23, mainly because they’ve also scored only 23 goals instead of 33. Their finishing has been pretty atrocious, from Andre Gray missing an open goal five yards out against Sheffield United, to not one, but two players missing the ball altogether when presented with great chances against Liverpool.
12. Arsenal (10th)
It’s at this point you realise they’re also tied for 13th in the table that matters, which some might call suboptimal.
11. Brighton (15th)
What Graham Potter would give to have eight extra points at this stage. While Brighton are faring better than expected, they’re still only two points above the relegation zone, and yet they could be virtually home and dry if only they’d put away good chances in a handful of games.
They’ve scored 30 goals this season from a total xG of 35. That isn’t a horrendous shortfall, but it’s seen them drop points in 1-1 draws with West Ham, Aston Villa and arch-rivals Crystal Palace. Some better finishing and the Seagulls could be comfortably mid-table. Arguably, Potter deserves even more praise than he’s already getting.
10. Tottenham (5th)
Although a five-place difference looks dramatic, it’s only three points. We can tell you exactly which three.
Shot data suggests Jose Mourinho hasn’t overseen much improvement yet at Spurs. He has, however, overseen a 2-0 win over Manchester City despite an xG scoreline of 0.4-3.2 to City. While the Cityzens (ugh) missed chance after chance after chance, all from positions with high-value xG, debutant Steven Bergwijn scored a stunning trap-and-volley. It’d be wrong to call the win a fluke – Bergwijn and fellow scorer Son Heung-min are Tottenham players, not some random deus ex machina – but it wasn’t exactly down to tactical brilliance on Mourinho’s part.
9. Sheffield United (6th)
This should be considered positively, in spite of the three-place difference. In fact, based on chance creation, the Blades should really have scored half a dozen more goals (yes, David McGoldrick, may well you look sheepish). Sheffield United are where they are on merit. They are challenging for Europe on merit. They’re imprinting a certain unique style on the division and they deserve to be treated as a good Premier League outfit: on merit.
That said, Dean Henderson conceding only 23 goals from 32 xGA helps. It also implies he’s the second-most effective goalkeeper in the Premier League. Could a currently-uncapped keeper be England’s No.1 at Euro 2020?
8. Southampton (13th)
Now this is a surprise. After riding out a storm that was part misfortune, part masochism – they lost to Burnley, Liverpool and Bournemouth despite winning the xG battle in all three matches – Southampton now look genuinely quite good.
Unfortunately, while Danny Ings has been very reliable in front of goal, the same can’t be said of his team-mates. The ‘Such Deep’ Long-Ings partnership has seen Ings score 14 goals from 10.21 xG, which makes him the Premier League’s fourth-best finisher this season after Harry Wilson, Jamie Vardy and, uh, Jonjo Shelvey. Shane Long has one goal from 3.26 xG, because he’s Shane Long and he’s preternaturally incapable of scoring.
7. Everton (9th)
Without wanting to get too technical, Everton find it easy to concede and hard to score. Their shooting positions should have resulted in half a dozen more goals, and at the other end, they’ve shipped five more than their xGA. However, things are looking brighter under Carlo Ancelotti since the club’s Christmas kerfuffle.
6. Wolves (8th)
This comes largely down to a defence that’s conceding more goals than it should be, because Wolves’ xGA is the fifth-lowest in the Premier League. At times, they’ve been unfortunate; other times, they’ve allowed an opponent – including Newcastle, twice – to get a result on the back of very few chances. In attack, Diogo Jota has three goals from a total 7.13 xG, which may help to explain why he isn’t the automatic pick that he was last season, when he was first moved into Wolves’ front line.
Even so, Nuno deserves yet more recognition for taking Wolves into a lofty position despite the Europa League fatiguing an inexplicably small squad.
5. Leicester (3rd)
After a run of teams whose xPts tallies suggest they could be doing better, we have Leicester, who appear to be overperforming. The Foxes have outscored their xG by 12 goals – nobody else can better seven. They’re riding the crest of a wave generated in part by Jamie Vardy’s excellent finishing and Kasper Schmeichel unexpectedly pulling off the season of his career. Vardy’s xG is the same as Chris Wood’s. His goal return is marginally better.
Leicester are still a very good side, though. Our eyes can tell us that, and also xG data for individual matches suggests they haven’t exactly been stealing wins on a regular basis. They’ve been the better side in nearly all of their 15 victories, although they were perhaps lucky to sneak past Southampton by such a narrow 9-0 scoreline.
4. Manchester United (7th)
Even with the revelation that’s still to come, this may be the biggest challenge for non-believers: according to the data, this Manchester United side… aren’t… bad? At least, their defence isn’t. That’s ironic in itself, because when we did this in 2018, David de Gea was single-handedly saving United by single-handedly saving goalbound shots – now, however, not so much. If he was, things could be different, as United have the league’s third-lowest xGA.
They could also do with some goalscoring help from their midfielders. The forwards are putting away chances well enough, but Jesse Lingard, Andreas Pereira, Juan Mata, Paul Pogba and Fred have contributed one goal between them this season from a combined xG of 7.5.
3. Chelsea (4th)
2. Liverpool (1st)
Whoa, whoa, whoa. Hold on now. Let’s try that again. What?
Yes, Jurgen Klopp’s 2019/20 Liverpool, statistically the best club team of all time, anywhere, apparently shouldn’t even be top of a table they’re currently leading by 22 points. This is the kind of thing that turns people against xG statistics, so let’s at least try to explain it.
It was always likely that a team with 73 points from 75 available may have been overperforming their shot data. Liverpool have been superb but nobody would’ve expected them to drop only two points all season – well, except Mark Lawrenson, who has been predicting this for years. Even so, it feels like overkill to say Liverpool would be expected to have dropped another 21. They clearly belong at the top of the league.
But their low xPts calculation isn’t based on nothing. In the first half of the season, Liverpool recorded several narrow wins in games of fine margins. Stattos and data-haters alike might agree that Klopp & Co. rode their luck to win at home to Watford and away at Chelsea, Sheffield United, Crystal Palace and Wolves. Furthermore, they kept only two clean sheets in their first 15 matches, and that’s a very rare way to win league titles.
You may scoff. Hell, we’re sceptical ourselves. And this is why: when you take 97% of the points on offer, and a metric dictates that you’ll virtually never take all three points from a game, the statistics <i>will</i> say your points tally is inflated. Even when Liverpool dismantled Leicester on Boxing Day, they were given 2.98 xPts instead of 3, because one time in 100 that identical match ends in a draw – and those game-by-game discrepancies add up.
However, to suggest that Liverpool have been a little fortunate as well as a lot fantastic isn’t totally absurd, and neither does it undermine their magnificent achievement. It’s a freak season. They’re still phenomenal.
1. Manchester City (2nd)
Top by five points, eh? When they’re actually 22pts behind Liverpool? What’s gone wrong?
One issue is that several City players’ finishing has gone to pot. Although Aguero is still an alien, averaging less than an hour per goal or assist, the list of players who are underperforming their xG includes Gabriel Jesus, Raheem Sterling, David Silva and Ilkay Gundogan (while Aguero betters his xG nearly every single season, Jesus never has). They’ve also missed Leroy Sané in a big way, with Sterling being forced wider in the German’s absence just as he was threatening to become a truly elite goalscorer. And they keep arsing up penalties.
The bigger picture is City’s sudden inability to win games they dominate. Oddly enough, that wasn’t a factor when they were racking up 198 points across two league seasons. In 2019/20, though, it’s a real problem. Case in point: the aggregate xG score of their two matches against Tottenham was 6.1-0.6 to City – and the aggregate score was 2-4. Pep Guardiola had better hope they don’t face Spurs in the Champions League again.
While you’re here, why not take advantage of our brilliant subscribers’ offer? Get the game’s greatest stories and best journalism direct to your door for only £12.25 every three months – less than £3.80 per issue! Save money with a Direct Debit today