Football’s Longest Partnerships, Ranked

Alessandro Costacurta and Paolo Maldini played together at AC Milan for 21 years. Not a stat anyone publishes, because it needs two careers lined up against each other rather than one career read down the page. We have 739 players with verified, dated club spells, which yields 11,423 real partnerships — so here is the leaderboard, and then a question worth asking of it.

The longest partnerships in the database

In the chart above, each bar spans the years two players actually overlapped at the same club, with the end caps standing for the two players. The top of that list is a run of one-club defences and midfields that stopped existing some time around 2010. Alessandro Costacurta and Paolo Maldini anchor it at 21 years, which is longer than most professional careers, let alone most partnerships inside one.

The pairs that moved together

17 of the 150 published partnerships happened at more than one club — the players did not merely share a dressing room, they kept sharing one after a transfer. The extreme case is Frank de Boer and Ronald de Boer, who turn up together at 4 different clubs across two continents (Barcelona, Al-Rayyan, Al-Shamal Sports Club, Ajax) — though being twin brothers is a confound no statistical model is going to improve on.

So: do long partnerships pair up by star sign?

This is the obvious question to ask of a table of durable two-person relationships, and the internet is full of confident answers to it. Ours is a flat no, and the interesting part is the yardstick it takes to say so honestly.

The tempting comparison is a flat zodiac: twelve signs, so two people should share one about 8.3% of the time. But our earlier look at footballer birth dates showed the roster is not evenly spread — the January academy cutoff bunches players into the first quarter of the year. Two players drawn at random from this pool already share a sign more often than a flat zodiac predicts. Measure against 1-in-12 and you would find an “effect” that is nothing but the selection bias, restated.

So the baseline below is computed exhaustively over every possible pair of the players who appear in these partnerships — 6,670 combinations, not a sample. That is what chance actually looks like here.

The fifty longest partnerships share a sun sign 10.0% of the time. Chance for this pool is 8.5%. Across all 150 published pairs it is 8.7%. Element matching lands under the baseline rather than over it. There is no gap here to explain, in either direction.

Why that null result is not the debunking it looks like

It would be easy to stop there and call it a takedown of astrological compatibility. That would be sloppy, because it attacks a claim serious practitioners do not make.

Sun-sign matching — the “Leo and Aquarius are a bad match” genre of magazine column — is not what relationship astrology actually consists of. The technique for comparing two people is synastry: you draw both natal charts and read the angles between one person’s planets and the other’s. It looks at Venus against Mars, Moon against Moon, where one chart’s planets fall across the other’s houses. The sun is one body out of ten, and the exact minute and city of birth change the reading completely.

In other words, the thing our data can test is the one input astrologers themselves treat as the coarsest — and our null result is what both a sceptic and a synastry practitioner would predict, for different reasons. Anyone curious what the actual technique looks like on two real people can run one through a synastry chart calculator and see how little of it the sun sign accounts for. What our database holds — a date, no time, no city — cannot get anywhere near that, which is the honest end of this line of enquiry rather than a verdict on it.

What the data actually explains

There is a real pattern in the table, and it has nothing to do with birth dates. Every partnership at the top is a defender, a goalkeeper or a holding midfielder at a club with an academy and no financial reason to sell. Long partnerships are a product of squad stability and position: centre-backs and keepers stay, forwards get sold. The zodiac column is the decoration; the club column is the explanation.

How this was built

Partnership lengths come from Wikidata career spells, matched on club Q-id rather than club name, so two clubs with similar names cannot be conflated. Two players count as teammates for a year only where their spells genuinely overlap; sharing nothing but a transfer year does not count. National-team, regional-representative and reserve-side entries are excluded, which is why the Basque Country XI does not appear as a sixteen-year partnership between two Athletic Club players.

One caveat worth stating plainly: this is the overlap our database can verify, not a claim about football history. A partnership at a club we do not track, or between two players we do not carry, is simply absent. The ranking is of 11,423 pairs we can check, and nothing more.

Play with the same data

Teammate overlap is the engine behind several of our daily puzzles. Teammate Bridge asks you to find the player connecting two others, and Six Degrees stretches the same idea into a chain. If you would rather test the career side of it, Career Path reveals a player one club at a time.