Before we begin, a quick note: Today’s piece involves the kind of research that normally goes into work I do elsewhere. But I decided to run it here at No Grass In The Clouds, something I hope to do more frequently in order to make this a self-sustaining newsletter. If you’d like to see more of this — more research, more reporting — then please consider becoming a paid subscriber. It’s $7/month or $70/year, and you get double your current fill of No Grass in the Clouds.

Also, this piece is for Alex S., who donated to COVID relief and requested that I write about HMS.

Heung-min Son isn’t gonna keep this up. 

Through 11 Premier League games, the Korean superstar has found the back of the net with 47.6 percent of his shots. He’s averaging about two non-penalty shots per match; extrapolate that out to 38 games, and he’s on pace for 76 shots. In Europe’s Big Five leagues since 2013-14, 710 players have attempted at least 70 shots in a season. The average shot-conversion rate among those players is 12.6 percent. The high-water mark is Antoine Griezmann’s 2014-15 season with Atletico Madrid, when he scored 22 goals on 78 attempts (28.2 percent), trailed closely by Luis Suarez’s 37 goals on 132 shots (28.0 percent) for Barcelona the following year. 

Son’s conversion rate over that same stretch is 16.2 percent, and his career-high was 17.5 in 2016-17. Plus, as you might’ve caught on Sunday, it’s not like Son is suddenly taking the kind of high-value shots that might, say, lead one to convert his chances at nearly double the highest rate we’ve seen over the past eight years. Based on the location of the attempts plus a number of other factors, he’s taking shots worth 0.34 expected goals per 90 minutes -- just slightly above his overall average of 0.32 since 2013. He’s not getting tap in after tap in; no, he’s throwing boulders into the upper corner.

Son won’t keep scoring with every other shot, and Son is one of the best finishers in the world; these two things can also be true at once. You don’t just score that many chances from that many shots without some kind of special skill. It’s not all luck, it’s not all bad goalkeeping. Heung-min Son is better at kicking a soccer ball than maybe anyone in the world other than Lionel Messi. Last year in a piece I wrote for ESPN, I found that Messi had added about 33 goals with his finishing since 2014. No one else is above 13. Messi is the best creator, facilitator, dribbler, and shot-getter of his generation -- and he’s better at shooting than everyone else, too.

But who might be second? And could it be Son?

Shot Conversion

Let’s start here. The best shooter in the modern era of European football is ... Tim Howard?

Since 2008, four players have converted 100 percent of their shots: Howard, Valencia center back Hugo Guillamon, and Howard’s fellow keepers Ali Ahamada of Toulouse and Alberto Brignoli of Torino:

But for our purposes, we’re gonna start the cut off in 2013 and we’ll only include players who’ve attempted at least 70 shots since then. The leader in shot-conversion is the gigantic Dutch journeyman Bas Dost, who, at 6-foot-5, has converted 27.7 percent of his shots into goals. Second is Dost’s polar opposite, Jadon Sancho, the Borussia Dortmund wide forward who is A) a legitimate superstar, and B) only 20 years old, and C) scoring on 25.4 percent of his shots. Kylian Mbappe is fifth (22.7) and Luis Suarez is 15th (20.7), but the 12 other players in the top 15 are mostly poacher-types who get on the end of a relatively low number of chances in the penalty area but don’t really stand out in any other facets of the game. Think Miroslav Klose and Chicharito Hernandez. In other words, for the most part, the best attackers are not the players who convert the highest percentage of their chances. Son ranks 107th.

And while we’re here, spare a thought for Leganes defensive midfielder Ruben Perez, the only player in the dataset not to score a single goal. And it’s not like he’s barely scraping over the 70-shot threshold; the guy has taken 132 shots! A quick look at where he’s taking the shots from might help to explain one none of ‘em have gone in:

The day after he finally scores should be a global tax holiday for everyone below a certain income level. Someone please alert the UN.

Goals Above Average

OK, back to the good shooters. Shot-conversion rate clearly states the obvious: what players are turning the highest percentage of their shots into goals? But it doesn’t take into account the location of the shot and the volume of attempts. That’s where xG comes in. Stats Perform defines the stat as follows: “Expected Goals measures the quality of a shot based on several variables such as assist type, shot angle and distance from goal, whether it was a headed shot and whether it was defined as a big chance.” While nowhere near perfect, the xG models are easily the best metric we have to both predict who’s going to score goals in the future and also understand the nature of how goals come about. The main finding from the development of xG was that the majority of a player’s goal-scoring could be explained by a combination of how many shots he took and how good those shots were. Obvious in hindsight -- like so many analytical insights throughout the course of spots history.

Bas Dost, then, is not a better striker or goal-scorer than Robert Lewandowski because he converts more of his shots. No, Lewandowski gets on the end of so many more shots that a lower conversion rate is more than made up for by a way higher sum of attempts on goal. On top of that, were Bas Dost able to get on the end of as many chances as Lewandowski, he almost definitely wouldn’t still be scoring with more than 25 percent of his shots.

However, an ability to convert your attempts at an above average rate doesmatter on the margins; it might take a given player from 22 to 26 goals or, say, five to eight in a given year. So, another, slightly less foggy window into finishing could be just comparing goals to expected goals and seeing who has more. Stats Perform calls it “Goals Above Average”.  Here’s the top 10:

While this may be painful reading for Barcelona and Real Madrid and Argentina fans alike, it’s clearly a more useful way of looking into what we’re looking for: players who both take a lot of shots and convert a lot of shots. These players are all great goal-scorers, legit stars, and staples for their national teams -- to varying degrees. This, of course, favors players who have had more seasons to rack up even more overperformance, but maybe it also weeds out the players who couldn’t sustain their finishing from year to year.

If we look at goals above average per shot, then Sancho is atop the list along with, strangely, Inter Milan center back Stefan de Vrij, at 0.08 goals above average per shot. Plotting both stats on the same graphs paints Son in an even better light, too. He’s circled in red.

If we accept that Messi, Kane, Son, Griezmann, and Suarez are five finishers in a class of their own -- and this chart suggests as much -- Son is actually doing the most with each individual shot among that select quintet. He’s added 0.06 goals above average with each attempt since 2013 -- meaning, the average Son shot over the past eight years has been six-percent more likely to go in when compared to what the xG model says -- while none of the other four are above 0.05.

That’s obviously Messi all the way to the right, and then it’s Martin Brahtwaite all the way to the left. Go ... Barca?

Shooting Goals Added

However, one of the things that’s not being included in any of this: the goalkeeper. If all of the other factors are equal -- pass type, shot location, body-part used to take the shot, etc. -- and one player picks out a corner, while another player hits it down the middle, they’ll receive the same xG credit. And if the former has his shot tipped around the post by a great save, while the latter sees his shot squirt through the opposing keeper’s legs, the guy who took the worse shot will get credit for finishing the chance, while the other one gets a big ol’ zero.

To deal with this, Stats Perform has a model that takes the location of the shot on the goal-frame into account, too:

The Expected Goals on Target model is built on historical on-target shots and includes the original xG of the shot but also the goalmouth location where the shot ended up. It gives more credit to shots that end up in the corners compared with shots that go straight down the middle of the goal. This model is only for on-target shots given that if you don’t get your shot on target, there’s a 0% chance that it will result in a goal.

You can then compare the xGOT to the xG to see which players are adding value to their shots by picking out corners and which ones are extinguishing great chances by not hitting the goal frame. Using xGOT also eliminates the opposing keeper’s performance as a variable. By this metric -- xGOT minus xG, or “Shooting Goals Added” --  Messi looks even more like some kind of weirdly specific joke meant to break the aggregate-power of this algorithm, while the rest of the above-average finishers start to muddle together.

Once again, though, Son sits somewhere within the top group after Messi. He and Suarez are the only two players that fit into both of the elite non-Messi tiers, for Shooting Goals Added and Goals Above Average. Here’s every shot Son’s taken since 2013 in domestic play, sized by the xG value of the individual attempt.

Now, while SGA removes the opposing keeper from the equation, it also eliminates for some of the other kinds of variables that would cause a player to consistently score more goals than expected: fewer defenders in the way of the shot, a quick release, the ability to shoot with both feet, and, well, the ability to kick the ball really freaking hard. The latter three and specifically the two-footedness, apply to Son as much as anyone.

Given all that, I like the very unscientific method of combining both the stats together to make a decision. Due to the blindspots in the models, the limited number of shots taken by any player in any season, and just generally the fluid nature of the sport, all attempts at quantifying “finishing skill” have a high degree of uncertainty inherent within them, so our eyeball-the-two-metrics approach fits just fine for now. And given that Suarez and Son are the only two players in the top five of both categories not named Messi, the best-finisher designation comes down to the two of them. And given that Son is ahead of Suarez on a per-shot basis in both stats -- 0.03 SGA/per shot, compared to 0.02 SGA/per shot for Suarez -- we’re gonna give him the nod. 

So, although Son is not going to score with every other shot for the rest of this season, the goal on Sunday and any of the other nine so far this season weren’t flukes, either. The next time you’re watching a Spurs match and Son winds up to shoot from outside the box or from a tight angle inside it, that specific shot will still be more likely to miss than turn into a goal. But the sheer fact that he’s the one shooting it? That means it still has a better chance of going in than that exact same shot, in that exact same position, with that exact same foot, taken by just about any other player in the entire world.

All data via Stats Perform.