Stay up to date with notifications from The Independent

Notifications can be managed in browser preferences.

How long can Son Heung-min keep out-running his own xG?

Data suggests that a player’s huge overperformance against expected goals catches up with them eventually, so are Son’s Europe-leading finishing stats destined to end or is he an extraordinary outlier?

Lawrence Ostlere
Thursday 07 January 2021 16:02 GMT
Son Heung-min has been outperforming his expected goals for five seasons
Son Heung-min has been outperforming his expected goals for five seasons (Getty/Independent)

If you’ve been following the Premier League this season then you will have taken notice of Son Heung-min, and specifically his finishing. Every time the ball comes near him it seems to propel itself into the net a couple of seconds later as if by magnetic force, and what appeared like a hot streak at first has continued to burn brightly.

His two goals in recent days, against Leeds in the league and Brentford in the cup, were both typical examples, swept beyond the goalkeeper with an economical precision – and precision is the word. Son has taken 28 shots in total this season, which ranks 30th in the Premier League. Yet he’s aimed 17 on target, which ranks ninth, and he’s scored 12 goals, which ranks second.

So far so impressive, but scratch under the surface and it is part of a genuinely remarkable wider trend. To gauge a more detailed measure of Son’s finishing, we need to compare his actual goals to his expected goals (xG), which rates the quality of each chance. A host of variables go into xG models like a shot’s location and the type of assist, and it’s compared to thousands of historically similar shots in the Premier League to produce a rating for a chance’s quality, where 1 is a certain goal and 0 is impossible to score. A tap-in might be rated 0.96 while a 20-yard shot might be 0.12.

This season Son’s chances in the Premier League have a total value worth 5.39 expected goals. He has scored 12 times, remember, so he has needed less than half the quality of chances of an average Premier League player to score each goal. It's a phenomenal output, and that is not the end of the story. Half a season is a relatively small sample size, yet when you compare Son’s goals against his xG for the past five seasons he outperforms expectations every time, so much so that he is in a league of his own – way clear of Lionel Messi, Cristiano Ronaldo or any elite forward you could name.

The table below puts his unique numbers into context. It includes every player in Europe’s top-five leagues to have scored at least 10 goals this season (bar three players with limited historical data: Bruno Fernandes, Reims’ Boulaye Dia and Dortmund’s Erling Haaland). It also includes a few renowned goalscorers for comparison like Sergio Aguero, Lionel Messi and Neymar to make up 20 of Europe’s best forwards over the past five seasons.

The results might raise some eyebrows. The conversation rates of Robert Lewandowski and Cristiano Ronaldo, two of the best players on the planet, appear distinctly middle of the road. Raheem Sterling may not be the wasteful finisher some paint him as. Karim Benzema probably is. And smiling down on the rest is Son Heung-min, the 27-year-old South Korean who has outscored his expected goals by 44 per cent, with 61 scored from chances worth 42.4 xG.

It is important to understand just how statistically freakish this is.

First, it may be surprising to know that football analysts don’t give much credence to ‘finishing’ as a skill at all. Historical data indicates that in Europe’s top leagues, the differences in finishing ability between professional footballers are minimal. Yes, players and teams will often enjoy hot streaks which defy their expected goals, but the theory is that in the fullness of time everyone will regress to the mean.

Compare for instance Ronaldo’s finishing with a run-of-the-mill Premier League striker, like Burnley’s Ashley Barnes. Over the past five seasons Barnes has scored 34 goals from chances adding up to an expected goals value of 34.63, meaning his chance conversation rate is about average. In the same period Ronaldo has scored 117 goals from chances worth a total xG value of 116.22, meaning his chance conversion rate is also bang on average. Finishing, it seems, is not what separates the great and the good. What makes Ronaldo – and all of the world’s best forwards – exceptional is not that they convert difficult chances but that they find so many of them. Ronaldo’s mastery of movement, his strength and leap, his timing: this is the armoury with which he defeats defenders, cultivating a sheer volume of high-quality chances which he converts so consistently into goals.

Given enough time, most teams and most players find themselves tracking roughly in line with expected goals, just like Ronaldo and Barnes, which is precisely why the model works in helping to predict what might happen next. Liverpool were outperforming their xG this season but more recently (and predictably) they have regressed towards the mean, while the reverse has occurred at Arsenal. Analysts don’t always agree on whether these bursts of finishing prowess or profligacy should be attributed to abstract notions of luck, or more tangible fluctuations in factors like confidence and motivation, but they do agree that such spells rarely last.

There are always some exceptions, some players and teams who consistently outperform or underperform xG by small margins. An example last season was Kepa Arrizabalaga, the Chelsea goalkeeper who conceded more than his expected goals-against tally. His replacement, Edouard Mendy, has proved close to average and a noticeable improvement. At the other end of the pitch the data shows how Lionel Messi, Harry Kane and Ciro Immobile all track above their expected goals tally by around 20 per cent, considered the typical outer limit of any boost from elite finishing.

Which brings us back to Son, who doesn’t fit into any football analyst’s conventional thinking. With 44 per cent more goals than could be expected over the past five seasons, he has resisted xG’s gravitational pull to a degree that cannot simply be explained away by natural variance or luck.

There are some theories to explain Son as an outlier, beyond simply being a finishing maestro. The five-season sample size is big but he has not taken nearly as many shots as others on the list in the time period, and more shots are likely to equal more regression. Then there is the type of shot: picture a Son Heung-min goal and you probably don't visualise a header, or a free-kick, or a penalty, or a bicycle-kick. That is not to say that Son isn't efficient in those departments too, but it could be argued that a disproportionate number of his chances – mostly one-on-ones and shots from around the box – fall in a comfort zone in which he excels, while most top forwards face more varied opportunities that test their strengths and weaknesses.

Son Heung-min has been outperforming his expected goals for some time (Independent)

Analysts The Independent spoke to had their own theories. “It’s difficult because xG models weren’t originally designed to evaluate a single player’s finishing ability,” says Mike L Goodman, the former managing editor of StatsBomb. It is possible that Son may therefore throw up quirks in the data. His natural talent with both feet, for example, might confound models which consider whether a shot was taken on the attacker’s stronger or weaker side. Son’s specialism on the counter-attack may also play into his favour statistically, because xG models struggle to interpret impeding defenders and so rate one-on-ones as a little more difficult than they might be. “The reality is just that it's very hard, from a pure stats and numbers approach, to get enough shots under your belt to say conclusively that somebody who has been a great finisher will continue to do that in the future,” says Goodman. “You can look at Son and say his numbers suggest that he's a really good finisher, absolutely, but it’s just about being careful saying what we can prove.”

Goodman agrees, though, that in a game of mostly uniform standards of finishing, Son’s numbers are startling. And perhaps this is the point. In the heart of Big Data maybe there is little to choose between the finishing prowess of one footballer and another, but on the fringes of numbers, at the spreadsheet’s poles, there might just be something or someone worth discovering. Son will soon regress and probably won’t finish 2020-21 with more than double his expected goals, as he’s currently tracking, but the long-term trends suggest he has mastered the art of finishing to an efficiency above even greats like Messi and Ronaldo, and it shows no sign of tailing off.

It is worth noting that, as a team, Tottenham have been performing above their xG this season too, and that after Son they also have No2 on the list of Europe’s top finishers. So, in a season which already defies convention in so many ways, perhaps Son and Kane will continue to resist xG’s pull towards the mean for a little while longer yet.

Join our commenting forum

Join thought-provoking conversations, follow other Independent readers and see their replies


Thank you for registering

Please refresh the page or navigate to another page on the site to be automatically logged inPlease refresh your browser to be logged in