Article by Ben Griffis
Knowing what leagues or clubs may be the most similar to each other is valuable for many reasons. One natural application for this information is player recruitment (as well as development, either loans or targets for sales).
In the absence of (relatively quality) data to grade leagues against each other, we need to have decent knowledge of each league. This is possible on a limited scale, and something I’ve personally tried to gain knowledge on, particularly in the “less popular” European leagues and in Asia. But I couldn’t even begin to tell you where any league from Africa or North America might rank among UEFA’s main leagues. So, when Opta introduced their Power Rankings, I saw it as a treasure trove of information.
Opta’s coverage includes well over 13,000 teams, ranging from New Zealand to Norway, Guam to England, and everywhere in between. Other publicly available ranking systems for global soccer are minuscule in comparison. 538 covers fewer than 700 teams, and clubelo exclusively covers the main leagues in Europe. I’m not going to tell you what to believe about the quality of Opta’s rankings, but I personally think it’s a great start and only seems to get better as time goes on. Further, having a sole source, and thus the same ranking method, for 13,000 clubs is important by itself. Once we start mixing ranking sources, we get into very dangerous territory.
Overview Of The Method
If you personally like Opta’s rankings, then this article can stand both as a proof-of-concept and as an article with interesting findings. If you do not like Opta’s rankings, then please use this only as a proof-of-concept of what can be done with even better ranking data. The caveat on everything in this article is that it it according to Opta’s Power Ranking data… What’s the old saying; don’t kill the messenger?
One of the ways we can use rankings is to create a system that tells us what leagues might be the most similar, quality-wise, as well as leagues that might be certain steps above or below a specific league. Then we can expand to see what clubs in other leagues might be comparable to a specific team, or what clubs a focal club may look at to recruit from. I’ve developed a first-draft system to address all of these.
I will not share everything around the method like I’ve done in other articles where I develop something. I want to share some initial results, but keep the exact method secret for now as I may have other plans for it or, in a perfect world, develop it further with a club/company 😉
But, what I can say is that the only input data at this time is each club’s rating and global rank. That’s it. Thus, I’m calling this “basic”, “rough”, and a “first draft”, since it lacks crucial information to make this even more useful like possession, number of attacks/shots in a match, formations, and so many other possible variables. This data is likely available through an API, but since I don’t work in a club or company full-time, I don’t have access to any APIs with this information easily available and the ability to keep it up-to-date.
Sample variables I create and use in this system include raw values of a club’s rating and global rank, the average rating of clubs in their league, the competitive landscape of the league, the club’s relative competitive standing in their league, and more. Basically, almost everything you can do with the ratings, I do do with the ratings. To an extent.
But again, it does lack key variables I’d prefer to have as inputs about what actually happens on the pitches in these leagues and clubs. If I had this data, there wouldn’t be drastic changes to, say, the top 10 most similar leagues to a specific league, but it would help define the “edges” so to speak. It would also help hone in on the most similar clubs in other leagues, and particularly what clubs might be interesting to look at for players, or what clubs you could send players to. Whereas right now, I’m confident about the outputs from a club quality standpoint, it can only be used in a general sense. Much more development is needed and in the future it would be a goal to be able to define a specific role for a player and see what clubs might play a tactical style that suits development of that role. But I digress.
Case Study: Turkish Süper Lig and Fatih Karagümrük
The proof-of-concept for this article will take the Süper Lig and Fatih Karagümrük as its case study. I’ll structure the example as follows:
- Most similar leagues to the Süper Lig
- Most similar clubs (in the most similar leagues) to Fatih Karagümrük
- Sample leagues in different tiers above and below the Süper Lig
- Sample clubs that might be realistic next steps for Fatih Karagümrük players
- Sample clubs that might be good options for Fatih Karagümrük to source players from (or send players on loan to develop at)
Most similar leagues to the Süper Lig
The math behind this section is straight forward. Using many variables, I calculate similarity scores for every league in Opta’s data. I have specific thresholds for “similarity”, and this also plays a role in developing the tiers above/below each league.
These are the 10 most similar leagues to the Süper Lig (using data at time of writing, of course):
| country | region | league |
|---|---|---|
| Hungary | Europe | NB I |
| Brazil | South America | Brazil Serie A |
| Norway | Europe | Eliteserien |
| Poland | Europe | Ekstraklasa |
| Cyprus | Europe | Cyprus 1. Division |
| Sweden | Europe | Allsvenskan |
| Spain | Europe | La Liga 2 |
| Croatia | Europe | HNL |
| Denmark | Europe | Danish Superliga |
| Russia | Europe | Russian Premier League |
Hungary’s top flight, NB I, comes out as the most similar, Brasileirão 2nd, and Eliteserien 3rd. It appears Scandinavia must be pretty comparable to the Süper Lig, as Norway, Sweden, and Denmark’s top flights are all in the top 10 (as well as passing the similarity threshold, mind you).
Below is an image visualizing the makeup of the Süper Lig and its 10 most similar leagues, sorted by the mean rating of the league. This way we can see this similarity in action.
Overall, we can see that these leagues all have fairly comparable ratings and makeups, but of course they’re not all the exact same. Insights we can take away at this stage are that there are plenty of leagues that could be relatively comparable to Turkey’s top flight.
Most similar clubs to Fatih Karagümrük
Now we can look at Fatih Karagümrük specifically and try to find clubs that might be comparable to it. For this, I only look at clubs in the top 10 most similar leagues. The way I think about it is, you can’t really be that similar of a club overall if you play in a drastically different league! However, even without this filter, the results do end up being pretty similar.
Below are the 15 most similar clubs to Fatih Karagümrük in the top 10 most similar leagues:
| team | region | country | league |
|---|---|---|---|
| Lokomotiv Moskva | Europe | Russia | Russian Premier League |
| Randers | Europe | Denmark | Danish Superliga |
| Fortaleza | South America | Brazil | Brazil Serie A |
| AIK | Europe | Sweden | Allsvenskan |
| Lillestrøm | Europe | Norway | Eliteserien |
| Omonia Nicosia | Europe | Cyprus | Cyprus 1. Division |
| Warta Poznań | Europe | Poland | Ekstraklasa |
| Dinamo Moskva | Europe | Russia | Russian Premier League |
| São Paulo | South America | Brazil | Brazil Serie A |
| Osijek | Europe | Croatia | HNL |
| Puskás | Europe | Hungary | NB I |
| Lokomotiva Zagreb | Europe | Croatia | HNL |
| Akhmat Grozny | Europe | Russia | Russian Premier League |
| Kecskeméti TE | Europe | Hungary | NB I |
| Las Palmas | Europe | Spain | La Liga 2 |
The most similar club to Fatih Karagümrük appears to be Lokomitiv Moscow in the Russian Premier League. Behind Loko are Randers, Fortaleza, and AIK all with pretty similar scores. Below are these 15 clubs visualized. Orange dots are the 15 clubs, and blue dots are the other clubs in each league.
We can see that while the clubs all have similar ratings, there’s more to it than just the ratings (which you start to see when you look at similar clubs below the top 15 most similar). Fatih Karagümrük is similar to the best teams in Hungary that aren’t Ferencváros, similar to mid-table Russian and Brazilian clubs, similar to promotion-playoff sides in Spain’s second tier…
Overall, I hope you’re starting to see the value in a method like this even just for basic things like defining similar leagues and clubs. Now let’s get into the fun parts of this article.
Leagues in tiers above & below the Süper Lig
I won’t get into any of the exact calculations, like I said above, but the next piece of this puzzle is to define steps for leagues above and below the Süper Lig. There are not a ton of leagues above the Süper Lig to begin with, so as expected we don’t find many that hit our criteria. The Premier League and Bundesliga are in a complete league of their own, by the way. I think it makes sense too, because it does take a very special player to be able to move to those leagues and see great success.
| country | region | league | league tier |
|---|---|---|---|
| Spain | Europe | La Liga | 4 Steps Up |
| Italy | Europe | Serie A | 4 Steps Up |
| France | Europe | Ligue 1 | 3 Steps Up |
| Belgium | Europe | Belgian Pro League | 1 Step Up |
| Netherlands | Europe | Eredivisie | 1 Step Up |
| Portugal | Europe | Primeira Liga | Same Tier |
| Slovenia | Europe | 1. SNL | 1 Step Down |
| Argentina | South America | Liga Professional Argentina | 1 Step Down |
| Saudi Arabia | Asia | Saudi Pro League | 1 Step Down |
| France | Europe | Ligue 2 | 2 Steps Down |
| Morocco | Africa | Botola 1 | 2 Steps Down |
| Finland | Europe | Veikkausliiga | 3 Steps Down |
| Kazakhstan | Europe | Kazakh Premier League | 4 Steps Down |
This is by no means exhaustive, and just includes a sample of leagues in other tiers. I included Portugal’s Primeira Liga because it’s the 11th-most similar league to the Süper Lig, and the only other league in the “same tier” category we have not seen (the other being the top 10 most similar).
Overall, I know several of these leagues pretty well, and they all pass my eye test. Do they pass yours? It’s also important to note that this is looking at the leagues as a whole, not just the top few teams. So while Benfica and Porto might be vastly better than plenty in the Belgian Pro League, the overall makeup of the league (and other factors in the calculations) compared to Belgium puts it a little below. This type of tier-generation is one area where having on-pitch variables would help better-define these lines.
It’s also important to note that these tiers are dynamic. They are all relative to the Turkish Süper Lig, rather than a static list of tiers I calculated beforehand. The calculation looks at relative differences in a league from the focal league, which happens to be the Süper Lig here.
Clubs that might be realistic next steps for Fatih Karagümrük players
Now we’re into the final two bullet points. First up is “upward-looking”, so clubs that could be decent landing points for Fatih Karagümrük players able to make that next step up. A key thing to note here is that this is general, and rough (as I’ve said before since there’s no tactical/on-pitch variables yet). We have to approach this without a specific player in mind, and think about it generally. For example, where could a good Fatih Karagümrük player go to next? Depending on how good, we could filter by clubs in specific “steps up”.
Again, I won’t share the nitty-gritty details, but can share the general theory. Basically, we’re looking for clubs in the same level leagues or better leagues where a player could take their next steps. For clubs in similar leagues, that should be at bare minimum a lateral move, but likely a move to a club who is relatively better than Fatih Karagümrük in their league. So for example, while Lokomotiv Moscow in the Russian Premier League is a similar club to Fatih Karagümrük, Krasnodar are better and they actually show up in this list.
As we move to higher and higher steps up from the Süper Lig and Fatih Karagümrük, we have to look at teams who might be around the same relative level in their league or a little worse. The theory behind this is that, generally speaking, you’re going to make a step up in a league, but down in terms of relative quality of your new team in that new league. The best players can move from a decent team in a decent league to a great team in a great league, but again, this is built with a general purpose in mind.
Another way to think about it is that the best clubs in Belgium rarely send players to the best clubs in England or Italy, etc. They’re sending players to clubs more around mid-table. Even Junya Ito, one of the best players in the Belgian league for a couple years (played at Genk, one of the best teams), moved to a mid-table Ligue 1 side, Reims.
So, with that all said, here the 15 exemplary clubs Fatih Karagümrük players might look at for that next step in their career. Depending on age/potential, some may be better suited for “Same Tier” clubs, others maybe for 3-4 steps up (subjective names for objective tiers discussed above).
| team | league | league tier |
|---|---|---|
| Las Palmas | La Liga 2 | Same Tier |
| Levante | La Liga 2 | Same Tier |
| Eibar | La Liga 2 | Same Tier |
| Atlético Mineiro | Brazil Serie A | Same Tier |
| Auxerre | Ligue 1 | 3 Steps Up |
| Granada | La Liga 2 | Same Tier |
| Deportivo Alavés | La Liga 2 | Same Tier |
| Standard Liège | Belgian Pro League | 1 Step Up |
| Kalmar | Allsvenskan | Same Tier |
| Krasnodar | Russian Premier League | Same Tier |
| Elfsborg | Allsvenskan | Same Tier |
| Cercle Brugge | Belgian Pro League | 1 Step Up |
| Cremonese | Serie A | 4 Steps Up |
| Sampdoria | Serie A | 4 Steps Up |
| Internacional | Brazil Serie A | Same Tier |
| Flamengo | Brazil Serie A | Same Tier |
| Elche | La Liga | 4 Steps Up |
| Adana Demirspor | Süper Lig | Focal League |
| İstanbul Başakşehir | Süper Lig | Focal League |
| Rosenborg | Eliteserien | Same Tier |
| AEK Larnaca | Cyprus 1. Division | Same Tier |
| Nordsjælland | Danish Superliga | Same Tier |
| Brest | Ligue 1 | 3 Steps Up |
I decided to share all 23 clubs that the algorithm output because I wanted to say that there are more than 23 potential destinations, but there are 23 that the algorithm thinks a general player at Fatih Karagümrük right now could move to and hit the ground running. Like I said earlier, this is just a general sample. I tried to code some element of realism into it, since realistically speaking a Fatih Karagümrük player right now (not any specific one there today, but just in general) probably wouldn’t be able to move to Marseille and hit the ground running. But Auxerre or Brest, however, would likely be more suited to their skills at this time. But again, this is just general and non-exhaustive.
We could also flip this on its head and look at these clubs as possible options for Fatih Karagümrük to buy players who are either on the decline and could move down to Fatih Karagümrük, or who are struggling to get minutes. These players might not be the best players in their current clubs/leagues, but might be able to perform well for Fatih Karagümrük.
Clubs that might be good options for Fatih Karagümrük to source players
The flip side of the better teams/tiers are lower teams & tiers that could be viable options to source players from. I have to note here that part of the similarity calculation is the competitive landscape of the league, so it will prize teams in leagues which are relatively similar to the Süper Lig than leagues which are not. The reasoning here is that players coming into the squad should be better-adapted to the makeup of the league. For instance, some leagues (like the A-League Men) have most teams at about the same level. Others, like the Scottish Premiership, have a couple teams insanely better than the rest, with most the rest being similar. Others, like La Liga 2, might have most teams about the same but with a few stragglers… The key thing to note is that there are many types of leagues, and this model does try to take that into account.
In a similar line of thought to moving up the tiers, moving down the tiers means we should be looking for teams in “worse” leagues that might be relatively better in their league than our focal team is in their own league. In the Süper Lig, Fatih Karagümrük are slightly better than the average team according to Opta. That means they might look at teams in the same tier as them but maybe average or a little worse than average (those players could see Fatih Karagümrük as a step up), or for clubs who are in the next step below the Süper Lig but a bit better than average in their league, or maybe for one of the best teams in a league a couple steps below the Süper Lig. And so on.
Below are the top 15 clubs that could be potential options for Fatih Karagümrük to look at.
| team | league | league tier |
|---|---|---|
| Lokomotiv Moskva | Russian Premier League | Same Tier |
| Konyaspor | Süper Lig | Focal League |
| Arouca | Primeira Liga | Same Tier |
| Randers | Danish Superliga | Same Tier |
| Fortaleza | Brazil Serie A | Same Tier |
| AIK | Allsvenskan | Same Tier |
| Lillestrøm | Eliteserien | Same Tier |
| Omonia Nicosia | Cyprus 1. Division | Same Tier |
| Warta Poznań | Ekstraklasa | Same Tier |
| Dinamo Moskva | Russian Premier League | Same Tier |
| São Paulo | Brazil Serie A | Same Tier |
| Sivasspor | Süper Lig | Focal League |
| Osijek | HNL | Same Tier |
| Austria Wien | Austrian Bundesliga | Same Tier |
| Puskás | NB I | Same Tier |
We can see some of these clubs were the clubs deemed most similar to Fatih Karagümrük, because I have more or less lateral moves built in, since that happens all the time.
If we filter to show the top 15 teams in leagues deemed to be at least a step below the Süper Lig, some fun leagues and teams show up. Players in these teams might look at Fatih Karagümrük as a step up and want to move to the Istanbul side.
| team | league | league tier |
|---|---|---|
| Sepsi | Romanian Liga I | 1 Step Down |
| Olimpia | Paraguay Division Profesional | 1 Step Down |
| Velež | Bosnian Premijer Liga | 1 Step Down |
| Racing Club | Liga Professional Argentina | 1 Step Down |
| Defensa Y Justicia | Liga Professional Argentina | 1 Step Down |
| PAOK | Greek Super League | 4 Steps Down |
| Luton Town | EFL Championship | 1 Step Down |
| Aucas | Ecuador Liga Pro | 1 Step Down |
| Middlesbrough | EFL Championship | 1 Step Down |
| Borac Banja Luka | Bosnian Premijer Liga | 1 Step Down |
| Olimpija | 1. SNL | 1 Step Down |
| Boca Juniors | Liga Professional Argentina | 1 Step Down |
| Cagliari | Serie B | 1 Step Down |
| LDU Quito | Ecuador Liga Pro | 1 Step Down |
| Raja Casablanca | Botola Pro | 2 Steps Down |
Overall, I think these clubs pass the eye test too. It’s easy to imagine a transfer of a good player from strong clubs in Argentina that aren’t River Plate to an upper-mid-table Turkish club. Similarly, it’s easy to see a transfer from a promotion-chasing Serie B side to Fatih Karagümrük, or from one of Slovenia or Morocco’s strongest teams.
Of course, in the future I would love to get more on-pitch variables into the mix to be able to better calibrate for teams possibly playing similar styles, which would mean players would be able to transition that much easier and faster to a club in a better league. But for now, with limited access to those variables, I still think this is a powerful tool and would be very useful.
Final Thoughts
The key to remember is that this is mainly a showcase of something I’m building. It’s a general tool with lots of important applications and will only be able to improve in the future. I also hope to add customization when possible, so that it can be less general and more specific for all sorts of clubs or player roles/ability/potential. Right now we have to come at this from a generalist perspective. Of course we know that if an incredible talent from Boca Juniors would likely move much higher than Fatih Karagümrük, so the use of this tool in its present state is for “good” players, whatever that entails.
I also want to say that the use of this tool is restricted to leagues below the Top 4 UEFA leagues of Premier League, Bundesliga, La Liga, and Serie A. The average clubs in those leagues are so far ahead of all other leagues that this can’t really account for such drastic differences. The similarity calculations and thresholds are set to a limit that I’ve come to through lots of testing, and while I could make specific changes for those 4 leagues, I haven’t yet. Since the global football landscape outside of those 4 leagues (and a lesser extent, Ligue 1 as well) is a fairly smooth ramp from leagues like Belgium, Netherlands, Portugal, Turkey etc down to the “worst” leagues with quality data, it was built with that in mind. And it appears to be as effective for Fatih Karagümrük as it is for Slovan Bratislava, FC Seoul, Columbus Crew, Cape Town City, Riga FC, 1860 Munich, Cesena, Brisbane Roar, Pakhtakor, and all clubs in between.
Finally, at some point I want to experiment with adding in geographic or cultural distance into the mix. We can’t deny that more transfers happen within regions than from any random corner of the world. Often this will be down to constraints (either organization, financial, or even legal/work permits), but possibly for convenience too, and naturally similar regions have similar cultures and possibly languages as well. This wouldn’t be paramount, but simply a fun set of variables to experiment with to add some more realism.
Stay tuned for more on this as I develop and fine-tune it! I’m not expecting to share the exact method and code, as I’m still thinking what I can do with this. If you happen to read this and work in recruitment and find this really interesting and have ideas for it/want to see a sample for your league, feel free to reach out. I’d love to see if this can be honed and customized to be used “live”, so to speak.
Cover Image by Hans Bruckmann from Pixabay
Café Tactiques 