How this works

Three numbers drive this whole site: Elo rating (how good a team is, right now), xWC (how many World Cups that quality should have produced, on average), and xCC (the same, for continental championships). Here's what actually goes into each.

Elo: a running score since 1872

Every team starts at 1500 the first time it appears. After each match, both teams' ratings move by the same amount in opposite directions — winner up, loser down, split down the middle on a draw — so the population average never drifts. Three things decide how big that move is:

• How surprising the result was. Beating a team you were expected to beat barely moves the needle. Beating a team rated 300 points above you moves it a lot.
• What was at stake. A World Cup match swings ratings 3x as hard as a friendly. Continental finals (Euro, Copa América, AFCON, Asian Cup, Gold Cup) sit in between, qualifiers a notch below that.
• The scoreline. A 1-0 counts as a normal win. A 4-0 counts for nearly twice as much — routs say more about the gap in quality than narrow wins do.

Home teams get a 100-point bump to their expected result before a match — not because they're actually stronger, but because home advantage is real and the model needs to account for it before judging the surprise of what happens. Matches at neutral venues (most World Cup knockout football) get none.

Teams with fewer than 30 matches on record are marked provisional — new footballing nations, or teams that mostly sat out qualification. Their ratings move faster and mean less until they've played enough to settle.

xWC: what should have happened, averaged over 100,000 tournaments

A rating tells you how good a team is. It doesn't tell you how many World Cups that should be worth — knockout football is small-sample and upsets happen. So instead of asking "who's best," we ask: if you replayed this exact tournament, with these exact teams at these exact ratings, a hundred thousand times, how many times would each team win it?

Each simulated match samples a scoreline from the two teams' ratings — better teams score more, worse teams score less, and the gap between them decides how lopsided that gets. Group stages, knockout brackets, and third-place qualification all replay exactly as the real tournament's format did. Level scores in a knockout tie go to a coin flip for the shootout — penalty-taking is close enough to 50/50 that we don't pretend the model can predict it from Elo.

Run that 100,000 times and average each team's title count, and you get xWC: the number of World Cups a team's quality was worth across its history, independent of how the ball actually bounced.

xCC: the same question, for continental football

xCC is expected continental cups — the identical calculation, run over a confederation's own championship instead of the World Cup. Same Elo ratings, same 100,000 replays, same group and knockout formats replayed exactly as each edition actually ran. That means every UEFA European Championship from 1960 to 2024, every African Cup of Nations from 1957 to 2025, every AFC Asian Cup from 1956 to 2023, every OFC Nations Cup from 1973 to 2024, every CONCACAF Championship/Gold Cup from 1963 to 2025, and every Copa América from 1916 to 2024; other confederations are a matter of adding their tournaments, not new maths.

A team's xWC and its xCC are separate numbers and are never added together, and neither are two confederations' xCC figures — a team's xCC usually comes from just one confederation. A team can have entries in more than one for one of two reasons: it changed confederations (Australia, moving from the OFC Nations Cup to the AFC Asian Cup), or it played as an invited guest in another confederation's tournament without ever leaving its own (the CONCACAF Gold Cup and Copa América have each repeatedly invited outside nations — Qatar, Saudi Arabia, and South Korea from the AFC, South Africa from CAF, and Mexico from CONCACAF guesting Copa América itself; Qatar's own Copa América appearance on top of its Gold Cup one makes it the only team with entries in three confederations at once). Either way the figures stay separate and never-summed. They answer different questions — how many World Cups a team's quality was worth, and how many continental titles it was worth — and a team can look robbed in one and lucky in the other.

The luck number

Titles won minus xWC. Positive means a team over-performed what their ratings deserved — they got lucky, or they were clutch, depending on your mood. Negative means the model thinks they were the better side across more tournaments than the trophy cabinet shows. It's not a moral judgment; it's just arithmetic on 23 tournaments' worth of small-sample knockout football.

Head-to-head predictions

Every head-to-head page predicts a result from the two teams' current ratings, so it answers "if they met today" rather than forecasting a scheduled fixture. The Elo gap sets each side's expected goals, and the win/draw/loss split and scoreline list are computed exactly from those two numbers — no simulation, so the figures don't wobble between builds.

Where it's weakest: the two scorelines are drawn independently, with no correction for the fact that real football produces more 0–0s and 1–1s than independence implies. Low-scoring draws are therefore under-predicted. The home-advantage figure is also a single global number, identical for every team, when in reality it varies by country, altitude and travel distance.

What this doesn't capture

• Squad news, injuries, form heading into a tournament — none of it. A rating is built entirely from past results; it doesn't know a team's star striker is injured this summer.
• Tactics, managers, or anything qualitative. Elo is a single number. Football isn't.

This is a model, not a verdict. It's built to be internally consistent and fun to argue with — not to settle the argument.

Frequently asked questions

What is xWC?
Expected World Cups — the number of World Cup titles a team's Elo rating says it should have won on average, based on 100,000 Monte Carlo replays of every tournament it entered.

Why doesn't a team's xWC match the number of World Cups they've actually won?
Knockout football is small-sample. A team can be the strongest side in a tournament and still lose to a hot goalkeeper on the day — or the reverse, and lift a trophy their rating didn't fully predict. xWC averages over 100,000 replays; the real tournament is one draw from that distribution.

What is xCC?
Expected continental cups — the same calculation as xWC, run over a confederation's own championship instead of the World Cup. That covers every UEFA European Championship from 1960 to 2024, every African Cup of Nations from 1957 to 2025, every AFC Asian Cup from 1956 to 2023, every OFC Nations Cup from 1973 to 2024, every CONCACAF Championship/Gold Cup from 1963 to 2025, and every Copa América from 1916 to 2024. A team's xWC and xCC are always shown separately and are never summed, and neither are xCC figures from two different confederations.

How is the Elo rating calculated?
A standard Elo update: the expected result comes from the rating gap between two teams, and the actual result moves each rating toward or away from that expectation. The size of the move scales with what was at stake — World Cup matches move ratings 3x as much as friendlies — and with how big the scoreline was.

Does a penalty shootout count as a loss in the model?
The simulation decides drawn knockout matches with a coin flip, the same as a real shootout. Penalty-taking skill isn't really captured by a rating built from 90 minutes of match results, so the model doesn't pretend to predict it.

Why do neutral-venue matches not get a home advantage boost?
Because there's no home team. The rating bump only applies when a team is actually playing in its own country — most World Cup knockout football is at neutral venues.

How far back does the data go?
Every international match on record since 1872. World Cups themselves only start in 1930, but the Elo ratings behind xWC are built from the full history.