FIFA Mode — talent is the floor

Eight tournaments, eight top-5 winners

Pre-tournament FIFA Men's World Ranking versus the eventual winner. The video-game edition in market each cycle is listed for context — note that FIFA 23 was the active edition during WC22 because the COVID-shifted November schedule put Qatar into the FIFA 23 release window rather than FIFA 22's. Exact top-5 ordering by edition isn't published in a single canonical place (fifaindex.com returned 403 in this scrape), so the load-bearing column below is simply whether the winner sat inside the top 5 going in.

Chemistry beats FIFA on WC22

On WC22's 31-team field, chemistry density predicted tournament finish better than paper talent: ρ(TCD, finish) = +0.704 versus ρ(FIFA-23 Overall, finish) = +0.548. Chemistry isn't a replacement for talent — the top-5 rule still holds — it's the multiplier on top. Argentina's tied- fifth FIFA Overall plus the nucleus chemistry around Messi (the densest knot on the field once you exclude their goalkeeper) won them the tournament. The two scatters below show where every WC22 side sat against both axes.

Over- and under-achievers

Three scatters, same 30 teams (Qatar excluded for missing FIFA-23 Overall). Gold ring marks the four semifinalists. Dot fill encodes WC22 finish (light = group exit, dark = winner). The first scatter is the direct view: FIFA Overall going in versus how far each team actually got. Teams above the diagonal trend overachieved; below it, underachieved. The two scatters below then show why — whether chemistry density (TCD) or shared club history was the multiplier (or the missing ingredient).

FIFA Overall vs WC finish  

The most direct over/under-performer view. X = published FIFA Overall in the game edition that shipped immediately before each World Cup (FIFA 23 for WC22, FIFA 19 for WC18, FIFA 15 for WC14, FIFA 11 for WC10, FIFA 07 for WC06). Y = stage reached. Toggle years on or off below to compare cohorts. Single-year view also surfaces the top over- and under-achievers. Germany, Belgium, Denmark (2022) sit high-FIFA, low-finish — classic underperformers. Morocco, Croatia, Argentina sit on or above their FIFA line — talent translated, with chemistry as the multiplier.

FIFA Overall data prior to 2022 scraped from fifaindex.com per WC-edition listing. Coverage isn't full 32-per-WC: fifaindex's national-team listing skips some federations (mostly AFC/CAF/CONCACAF), so several participants per WC are absent rather than imputed. See fifa_multi_year.json for per-year coverage and the explicit missing list.

FIFA-23 Overall vs Team Chemistry Density  

X = FIFA-23 published team Overall. Y = TCD (count of strong AW-JOI pairs on the squad). Germany sits high-FIFA, low-TCD — the quintessential underperformer (group-stage exit despite an 83 Overall). Morocco is the mirror image: low FIFA Overall, mid TCD, semifinal. Argentina over-performed — mid FIFA, high TCD, winner.

FIFA-23 Overall vs History Index  

Y = History Index (squad players with at least one current or former club-mate inside their own national squad). High-rated nations also tend to have high shared club history because their players concentrate at elite clubs — but not always. Saudi Arabia sits at the bottom of the FIFA axis but near the top of the History axis — almost the entire WC22 roster played in the Saudi Pro League, the most domestically-bonded squad in the field, and they famously beat Argentina in the group stage. The off-diagonal here is the recruitment story: who is built from a single club ecosystem versus who is sprinkled across Europe.

Does shared history correlate with winning?

The History Index needs no tracking data — just a squad list and each player's club. So we can run it on every World Cup back to 2006 and ask the most basic version of the chemistry question: do teams built from a shared club spine go further? For each squad we take the share of players whose tournament club has at least one team-mate in the same squad. The textbook spine teams score high — Spain 2010 (a Barça + Real Madrid core, 87%), Germany 2014 (seven Bayern players, 70%), Italy 2006 (a Juventus/Milan/Roma block, 78%) — and all three won. Toggle years below; hover any dot for its number.

X = shared-club history (% of squad with a club-mate). Y = stage reached. All 32 participants per tournament. The relationship is real but modest: pooled Spearman ρ = +0.185 (p = 0.019, n = 160), and it is positive in every individual World Cup (+0.30, +0.12, +0.04, +0.40, +0.05 for '06–'22). Shared history is a tailwind, not a guarantee — Croatia reached two finals (2018, 2022) with one of the most scattered squads in the field, the counter-example worth keeping honest. And the very top of the axis is its own confound: the highest scores are domestic-league monocultures — Qatar '22 (13 Al-Sadd players) and Saudi Arabia (an Al-Hilal/Al-Nassr block) sit near 90% yet exit at the group stage. That's shared history without elite talent — exactly why the slope is gentle rather than steep. Note this is the same-tournament-club definition (consistent across years and sourceable from public squad lists); the WC22 History Index in the scatters above is a richer multi-year club-overlap variant, so the two numbers differ.

Shared-club history vs WC finish  

Squad club lists scraped from Wikipedia's per-edition "FIFA World Cup squads" pages (2006, 2010, 2014, 2018, 2022) — each player's club at the tournament; clubs compared only within a squad. Finishes are the complete 32-team knockout results, cross-checked against fifa_multi_year.json. Built by research/scripts/build_history_index_multi_year.py; data in history_index_multi_year.json.

Tee-up: scoring WC26 squads as they're announced

The History Index is computable from a squad list and a club-history table — it doesn't need tracking data. As 2026 rosters firm up we can score every squad on shared-club density and combine it with EA Sports FC 26's published team Overalls to flag candidate over- and under-achievers before the tournament. The model has a backbone now: eight World Cups of top-5-winners and one tournament (WC22) where the same chemistry gradient that wins the WC22 sample is fully fit. WC26 is the first out-of-sample test.

The 2026 FIFA-Overall bundle is in data/wc26_rosters.json — 26 nations with EA FC 26 Overall, top-stars list, and a note flagging the 11 nations EA didn't license at launch (Belgium, Senegal, Colombia, etc.) for which the "team Overall" is an analyst proxy from individual player Overalls rather than EA's official aggregate. (Brazil was on that list at launch but rejoined via a multi-year CBF deal announced May 2026, so it's no longer flagged.) The tracking-based chemistry side stays empty until the ball is kicked — but the shared-club History Index needs nothing but a squad list, so we can already run it on the WC26 squads as they're announced (below).

Who has the most shared club history? — WC26  

The same % of squad whose club has a team-mate in the squad metric from the historical scatter, run on each WC26 squad as it's named. There's no finish to plot against yet — this is purely who walks in with the most shared club history. Two flavours sit at the top: genuine elite-club spines (Germany 7× Bayern, Spain 8× Barcelona) and domestic-league blocs (Czech Republic 10× Slavia Prague, Egypt 8× Al Ahly) — the WC22 chart shows the former tends to convert and the latter doesn't.

Pre-tournament FIFA Overall — WC26 candidate field

Top 12 by EA FC 26 published Overall. Asterisk = EA-unlicensed nation, Overall is an analyst proxy.

Shared club history vs EA FC 26 Overall — WC26  

Does shared club history track talent going into 2026? Among the WC26 teams we have a real EA FC 26 rating for, there's a modest positive link (Spearman ρ above) — elite squads concentrate their stars at a handful of big clubs (Spain 8× Barcelona, Germany 7× Bayern), which lifts both rating and shared history. It's the WC26 version of the WC22 "FIFA Overall vs History Index" scatter above.

We only plot teams with a real EA rating. EA does not publish an official team Overall for every WC26 qualifier (especially smaller nations), and we don't estimate the ones it omits — so some qualified teams aren't on the chart. The link is also softened by domestic-league blocs: Saudi Arabia and Iran, whose players cluster in one home league, post high shared history at mid ratings. (* marks an analyst proxy carried over from the source data — an aggregate of individual player Overalls, not an official EA team rating.)

Caveats, honestly

• Single-tournament chemistry sample. The WC22 chemistry-vs-finish correlation (ρ = +0.704) rests on 31 teams in one tournament. A second tournament (WC26) is the obvious out-of-sample test.
• FIFA editions diverge from real-world cycles. FIFA 23 was in market during WC22 because of the COVID-shifted November schedule; in earlier cycles the video-game edition didn't always sit cleanly inside the World Cup window.
• History Index is a coarse proxy for shared time. It counts players, not minutes — the pair-minutes view on the "Time → Chemistry" tab is the continuous version.
• Historical FIFA top-5 ordering not exhaustively verified. The table above uses the FIFA Men's World Ranking #1 going into each tournament (cleanly sourceable) rather than the video-game edition's team Overall #1, which was not scrapeable in this session (fifaindex.com 403). The load-bearing claim — every winner sat inside the top 5 — is robust to either definition.