Barcelona’s Lionel Messi (Reuters / Gustau Nacarino)

Great teams are outliers. Making a study of greatness is difficult for precisely that reason—the rules that have been made up to study everyday good and bad teams might not apply at the top level.

In European soccer, where few regulations prevent the formation of true dynasties, the study becomes a little bit easier. In Spain’s La Liga, Barcelona and Real Madrid have won nine of the last ten league titles and finished first and second eight times in the last decade. Their revenues and payrolls dwarf those of their competitors. If we want to study the “superteam” effect, this is the place to look.

One issue that has dogged soccer statistics is the presence of a “superteam” effect on shot conversion. Last season Barcelona scored 102 of its 618 non-penalty shots, a rate of about 17 percent. By contrast seventh-place Athletic Club of Bilbao scored only 8 percent of its non-penalty shots. This is a persistent pattern over the last several seasons. Real Madrid and Barcelona score about 15 percent of their shot attempts, and the rest of La Liga averages a conversion rate under 1-in-10.

Now, some of this effect has to do with the quality of chances created. Anyone who has watched Barcelona knows how the club’s tactics focus on creating high-quality chances near to goal. By a method called “expected goals,” I can create a statistical estimate of the quality of chances created based on factors like shot location, the type of assist pass, the type of attacking move that led to the shot, and so on. Indeed, as expected, Barcelona and Real Madrid create better chances than the other clubs in La Liga.

However, expected goals do not tell the whole story. While Barcelona’s expected shot conversion rate of 14 percent is more than half again better than the expected conversion rate for the rest of La Liga, it still falls a couple percentage points short of estimating Barca’s real goal-scoring rate. And the same is true of Real Madrid.

What causes this divergence? The simplest explanation is that this just reflects individual greatness. Whatever a team does together to create a scoring chance, if that chance falls to Lionel Messi rather than Jesús Navas the ensuing shot is more likely to be scored. So I set up a study.

I identified 50 players who have taken the field for Real Madrid or Barcelona in the last six seasons, who have also played elsewhere in the top five European leagues. The 50 include famous names like Gareth Bale, Luis Suárez and Zlatan Ibrahimovic, as well as lesser lights such as Ibrahim Afellay and Fernando Gago. I compared their shot conversion rates at Madrid and Barca to how they performed at other clubs. The sample constituted a little under 10,000 shot attempts, split between about 3,000 at Madrid or Barcelona and 7,000 for other clubs.

(To make the comparisons fair, I weighted each player’s statistics by the lower of their expected goals numbers. So Yaya Touré for instance only attempted 19 shots for 1.7 xG for Barcelona in the 2009-2010 season, but has taken over 300 shots for 25 xG since. His stats get a weight of only 1.7. By contrast David Villa, who sandwiched a run at Barcelona between stints at Valencia and Atletico Madrid, gets a weight of 27 based of his statistics: 30 expected goals for Barcelona, 27 for other clubs in Spain.)

The result is that these players, for the most part, outperformed expected goals wherever they played.

In the weighted average, these 50 players scored about 52 goals above expectation for Real Madrid or Barcelona, and they scored about 41 goals above expectation for other clubs. This result suggests that most—perhaps 75 to 80 percent—of the “superteam” effect is a relatively simple equation. Real and Barca spend large amounts of money to buy players who are excellent finishers as well as being elite in other aspects of the game, and there is nothing surprising about guys who cost over $50 million clinically dispatching a few extra chances. They do the same playing for Udinese, Arsenal or Valencia.

While this is broadly true at the population level, there is one striking exception. Gonzalo Higuaín has simply been a different kind of player since leaving Real Madrid for Napoli two years ago. He had been one of the most deadly finishers in La Liga, scoring 75 non-penalty goals from 263 shots, beating his expected goals (58 xG) by nearly 20. For Napoli Higuaín has scored 27 of 191 shots, exactly matching his 21 expected goals. Indeed, Higuaín accounts for a significant amount of the variation between “superteam” finishing and “other club” finishing. While greats like Gareth Bale, Luis Suárez and Zlatan Ibrahimovic have maintained finishing rates around their career averages, this is not true for every player equally.

What this suggests is that, as with everything in soccer, the underlying story is more complicated still.

On average it appears that individual player skill explains most of the “superteam” effect. However, there is important variance player to player. Surely this is in part random variation, but it probably makes sense that some players will see different interaction effects within great teams. Some will excel, some will stay the same, and some may struggle to fit into a reduced role. The best initial assumption should be that players will maintain their typical finishing rates at lesser or greater teams, but further scouting and analysis is necessary to draw conclusions at the individual level.