MLS gains ground in league rankings

Comparing domestic soccer league strength is hard. It is much more difficult to rank league strength in soccer than it is to compare teams within a league. There are few competitive matches across international boundaries and those that do exist only feature the best teams in a league. With the boundaries of the US, it's similar to comparing conferences in NCAA football and basketball. Both of those feature a large number of out-of-conference games. Even so, arguments about whether, say, the ACC or the Big 10 is a better basketball conference are endless.

But with the World Cup just on the horizon it may be possible to compare the strength of a league (say Major League Soccer) against other leagues. National Team coaches just selected the 23 players they consider most likely to help them earn the globe's greatest trophy. Those players came from places you'd expect (EPL, La Liga, Serie A, Bundesliga) and places with a bit less renown (MLS, the leagues in Azerbaijan, Ghana, Kuwait, South Africa). So you could develop a model where you rank leagues based on how many players they employ who are being chosen for the World Cup.

But it would be naive to just say that the league with the most players at the World Cup is the best. That fails to address a few things.

First, there are some pretty good national teams not going to to World Cup. Their players are also pretty good. So this exercise includes all nations going to the World Cup and every national team ranked by ELO as better than the worst nation going to the World Cup (Cameroon). That totals 55 national teams. Those that did not qualify are in friendly season, but those rosters can be experimental, instead we used the 23-man roster for their last World Cup Qualifying match in which qualification was still a possibility (except for the UAE, they were eliminated early so their recent Asia Cup Qualifying roster was used).

Second, having a player selected to play for a team like the UAE isn't as good an indicator of league strength as having a player selected to play for Brazil. But using ELO rating you can create a win % expectation. sidereal took the ELO rating for all 55 nations and created that win percentage as if the team was playing the median World Cup national team. That number is then granted for each player on the national team and then assigned to the league the player represents. For example, Julio Cesar earned MLS 84.60 for playing on Brazil, a national team whose ELO predicts would have an 84.6% chance of beating the median World Cup team. In contrast, Fabrice Olinga of Cameroon (the weakest team in the model by ELO) gives the Belgian league 22.5 points, as Cameroon is predicted to have only a 22.5% chance to beat the median World Cup team.

Some players are actually free agents right now. They were assigned to the national league structure where they were last playing. Other players have already signed new deals. They were placed with the new league structure. David Villa is odd (he's with NYCFC but on loan to A-League side Melbourne City), but since a total of 1,255 players were used that one quite peculiar case isn't important. The injured players that were already replaced could also be a factor, but again with the quantity of players used shouldn't be a large influence. There is also the issue of players on lower division teams. These were placed within the greater league structure as a recognition of those players and club teams trying to earn their way to the first division (so effectively the "English system" includes both the Premier League and the Championship). No lower-division players in the US/Canada system play on any of the top 55 ELO ranked teams.

Don Garber is very public with his goal to be a top tier league.

When this exercise was last done (in 2010) that seemed impossible. MLS was merely the 24th best in the world (with a slightly different model, though the ranking should be roughly comparable). Major League Soccer, by the current method, is now the 12th best league. The improvements do not come just from the strength of the United States national team. There is a wider variety of nations represented in MLS than in 2010. It used to be that just CONCACAF internationals and retirees came here. In 2014 Brazilians, Spaniards, Englishmen (that just miss out), Australians, Persians (Iran), and Irish all play in MLS and play roles on nations in the World Cup or of World Cup quality.

There's still work to do. Russia, Liga MX, Ukraine and Turkey are all non-traditional leagues that rate better. Portugal is still buying good players away from MLS and rated higher. The Eredivisie (a classic sell-on league that ) is still ahead. But the improvements are clear. Coming to MLS does not end an international career (at least not always).

Moving ahead of the Swiss league, the Greeks, and Belgium matters. Staying ahead of all of the oil-money powered Arab leagues is significant as well. They too are trying to use money to gain recognition globally.

For all its quirks MLS is doing some things right. It's not just about signing huge names. With Yedlin on the roster there is an example that the Academy system can work. Nations are looking to dual citizens playing within Major League Soccer and asking them to play for their non-US side. Peak age players are keeping their national team hopes alive (Fernandez with Dallas/Peru, who wasn't in this study, but was on their last friendly roster).

Top national league systems

Here's the full table of results, with the total number of points assigned by the model and the number of national teams (out of the 55 in the model) represented in the league.

Like all attempts at ranking national leagues this has flaws. Leagues with very good national teams but that do not import many foreigners are probably under-scored. For example, the Brazilian league neither employs many Brazil national team players (because they're all good enough to play in top leagues) nor many others (because it's not a destination league) so it gets few points. Also, domestic leagues that feed the bulk of middling national teams are probably overscored. For example, the entire 23-man roster of the United Arab Emirates national team plays in the UAE domestic league, and that large chunk of players inflates its score. But these aren't issues in the top 12 leagues.

And the model has advantages. Unlike those that count on international club competition it does not rate just the top teams. Unlike those that use friendlies it only includes competitive rosters. The values given for players are proportional to the quality of national team. So it's probably as good a model as you'll find that can handle every domestic league in the world.

Full data here