I had a theory: squads that have known each other longer have better chemistry, so they go further in the World Cup. I put every squad since 1930 on the lift and checked. The fault was not where I was looking.
vehicle488 national squads · 1930 to 2022 · 10,965 players
customer states"teams that have known each other longer win more"
suspectsthe wiring · chemistrythe mileage · caps
diagnosis · tl;drThe wiring is fine. Chemistry, measured properly, predicts nothing. The fault is on the odometer: past 40 caps average the engine strains, and no squad averaging 50+ has ever reached a semifinal. The wear sits in the midfield, the legs of the team.
parts usedall 22 World Cup squad lists · 149 continental tournament squads, 30,854 players · team strength rated from 49,500 internationals since 1872 · pass-by-pass data for 2018 and 2022
This is what my theory looks like as a machine. Every player is an atom, and every pair is joined by a spring: the more matches two players have shared, the shorter and stiffer their spring. If A has played 50 games with B and only 3 with C, the molecule shows it. My hunch was that tightly wired molecules go further. Drag the atoms. Pin them to their lines of the pitch to see where each squad carries its weight.
keeperdefensemidfieldattack· thick outline = likely XI · spring = est. matches shared
02The inspection
01The hunch checks out?
First test, the simple version of my theory. Every squad since 1930, placed by how long its players have shared the national team (average years since each player's first World Cup with the side) against where the team finished that summer. The line tilts up. Familiar squads finish higher, r = 0.20. Case closed?
02Subtract strength
Not so fast. Good teams keep their players, and good teams win anyway. So I rated every team's strength on the eve of each tournament, an Elo score built from 49,500 internationals since 1872, and asked a harder question: did the squad do better than its strength predicts? The tilt collapses, r = 0.03. Knowing each other longer moves nothing by itself.
03Check the wiring properly
Maybe "years together" was a lazy wire gauge. So I measured the real thing: for every pair in every squad, how many major tournaments they had actually been through together, World Cups plus Euros, Copa América, AFCON, Asian Cup and Gold Cup, 30,854 continental players scraped to close the circuit. Plot that true shared history against overperformance and you get this. A cloud with no slope. Chemistry, measured properly, predicts nothing. Not at squad level, not line by line.
04Switch dials: the redline
So the wiring is fine. Then I changed the gauge to mileage, average caps per player, and a shape finally appeared. Squads carrying 10 to 30 caps run slightly above expectation, and the 10 to 20 band reaches semifinals most often, 26%. Past 40 the engine strains. The ten squads above 50: zero semifinals, 17 points below expectation. Spain 2014 brought the most experienced squad ever measured and went home in the group stage.
05The age decoy
The textbook answer used to be "peak age is about 27". Swap the axis to age and a faint hill shows, but it is a decoy: control for mileage and the age effect drops to nothing, while the mileage redline survives controlling for age. It was never the birthdays. It is the odometer.
03Under the microscope
One last chance for chemistry. For 2018 and 2022 there is pass-by-pass data, so I could watch the wiring carry current: within each team, holding positions fixed, do pairs with more shared history actually pass to each other more?
128
matches, every game of 2018 + 2022
3,344
player pairs with 90+ shared minutes
+4%
extra passes per 90 between bonded pairs
0.2%
of pair-to-pair passing variance explained
A whisper. Who passes to whom is position and tactics, and familiarity adds almost nothing on top. The one place old bonds do show up: squads whose shared history is piled up front control games slightly less than their strength suggests. They grind, they do not dominate. My favorite footnote: Argentina was the most Messi-wired, bond-aligned team of both tournaments and won one of them, and even that alignment predicts nothing in general.
wiring · no fault found
04The engine room
If the fault is mileage, where does it sit? I split every squad into its lines and measured each line's mileage two independent ways, estimated matches shared (from caps) and plain age. Same asymmetry both times: worn defenses are fine. Worn midfields are not. Veteran defenses run on positioning. Midfields run on legs.
Each bar is the effect of one line's wear on results, holding team strength and the squad's overall profile fixed. Gray bars are indistinguishable from zero. The red ones are not, and they all live in the same place: the midfield, and the link between midfield and attack.
Spain is the whole story in one team. In 2010 the shared miles sat in the back, Casillas, Puyol, Ramos, and Spain were champions. By 2014 the same generation's weight had migrated up into the midfield and the front, exactly the zone the data says is fatal. Below, the same molecule four years apart. Watch where the springs thicken.
Spain 2010 champions
Spain 2014 group stage
05Reading the gauge
So the working instrument of this whole inspection is one number: average international caps per player on the day the tournament kicked off. The squad's mileage. Tap a squad to move the needle.
06Open workshop
Every squad since 1930. Hover a dot for its inspection sheet. Type a country to thread its whole service history through the cloud. Gold dots lifted the trophy, and notice where none of them live.
championsquad
07Written off · the 50 caps club
Every squad that ever entered a World Cup averaging 50 caps or more. Ten of ten failed to reach a semifinal.
08What actually moves the needle
Pre-tournament strength explains about a quarter of tournament results. Squad mileage adds a couple of points at the extremes. The rest is why we watch.
26%
71%
team strength (Elo)squad mileage · 3%football
chemistry, measured properly: 0%
09Also inspected · no fault found
Things everyone says matter, checked against the same strength control. All of them came back clean, which is a finding too.
✓Golden generations
Squads built around one debut cohort deliver exactly what their strength says. No more, no less.
✓Shootout experience
The more experienced side wins 19 of 35 World Cup shootouts. A coin flip.
✓Veteran keepers
No measurable premium for an experienced goalkeeper. Sorry, folk wisdom.
✓Club-mate blocks
Building the squad on one club's core, by line or overall, moves nothing once strength is held.
✓The age mix
Veterans-plus-kids barbells, one big generation, smooth pyramids: the blend does not matter.
✓Rising vs fading
Whether a squad's mileage is on the way up or down changes nothing. Only the level matters.
Inspection notes
Squads and caps at kickoff parsed from the 22 Wikipedia World Cup squad pages (10,965 players, 98.8% caps coverage) plus 149 continental tournament squad pages (30,854 players) for the measured shared-history test. Team strength: an Elo rating in the style of eloratings.net, computed from an open dataset of 49,500 international matches from 1872 to 2022, snapshotted on each tournament's opening day. Results measured as each team's points-per-game percentile within its own tournament, so eras with 13 teams and eras with 32 compare fairly. Overperformance is the distance between that result and what strength alone predicts. Passing analysis: StatsBomb open event data, every match of 2018 and 2022, pairs compared within team and position pair so adjacency does not fake chemistry.
The springs in the molecules are estimated matches shared, built from each pair's caps and the team's match calendar. It is a model, not a measurement; the measured shared-tournament test in act 03 is what the null verdict rests on. Honest caveats: only ten squads in history sit above 50 caps average, a small sample. A high caps average can also flag a federation with a small player pool that recycles its squad, and the strength control softens but cannot fully remove that. And nothing here is destiny: Argentina lifted the 2022 trophy at 34 caps average, just past the sweet spot.
One more inspection, read aloud…
When I'm not up to my elbows in a rabbit hole like this one, I build Koby. It turns ebooks you already own into audiobooks, and it can even read comics and describe charts out loud. It's free to try.
