Appendix. Methodology and supporting results

← Back to the article: 2022 vs 2026 FIFA World Cup: How is football (soccer) evolving?

This appendix explains where the data come from, how the two World Cups were compared, and which results support each statement in the text, the chart and Table 1.

A.1 Data

• FIFA Post-Match Summary Reports (PMSRs). All 64 matches of Qatar 2022 and all 104 matches of 2026, in PDF. They are the only source of match data: team statistics, phases of play, shots with their maps, player tables (in possession, out of possession and physical), defensive pressure, block height and length, goalkeeping, line-ups and match events.
• FIFA World Ranking before each tournament (October 2022 and April 2026), used to measure team strength.
• Temperature at kick-off for every match, from the Open-Meteo historical weather archive. Air-conditioned stadiums were set to 22 °C. In Qatar every stadium but one was air-conditioned; in 2026, three out of sixteen.
• Press reports, for the reason behind early substitutions in the last 16 of each tournament.
• Scientific literature, for the aerodynamics of the ball (see A.7).

Metrics are read according to FIFA’s official definitions for its Enhanced Football Intelligence (EFI) framework. FIFA does not publish the numerical thresholds behind its algorithms.

A.2 Data extraction and quality control

The data were extracted from the 168 reports with a custom Python program (pdfplumber library):

• Text and tables: read line by line and by the position of each word on the page.
• Graphics: event icons, shot maps, the goal-mouth chart and the goalkeeper intervention chart were read as vector objects, by shape, size and colour.
• Shot location: taken from the half-pitch shot map. The scale comes from the penalty area (40.32 m wide) and was checked against the penalty spot, which lands 10.99 m from goal.
• Shot placement: where each shot crossed the goal line, from the goal-mouth chart (7.32 × 2.44 m).
• Goalkeeper intervention: for each shot, whether the goalkeeper saved and held, parried, attempted a save and failed, or made no attempt.

Every match went through about 100 consistency checks against the report’s own totals: goals against the score, shots by outcome, the sum of players’ distances against the team total, substitutions on against substitutions off, among others. One error per tournament remained, with no effect on the results. 1,435 of 1,435 shots were located in 2022 and 2,570 of 2,571 in 2026.

A.3 How the two World Cups were compared

• Normalisation: counts are expressed per 90 minutes (value × 90 / match length), because some matches last 120 minutes. Percentages and averages are not normalised.
• Three comparisons for every metric (the three colours in the chart): (a) raw data, the direct difference between averages; (b) adjusted for team strength, a regression that accounts for the FIFA ranking of each team and of its opponent and for the stage of the tournament; (c) same 26 national teams: only the teams that played both World Cups, each compared with itself. Note that the adjustment does not weight goals: it estimates the 2026 effect while holding team and opponent strength constant. Figures in the text come from comparison (c) unless stated otherwise.
• Margin of error: 95% confidence interval. If it includes zero, the difference is not statistically significant.
• Many metrics at once: around 100 metrics were compared. By chance alone some would come out "significant", so a correction for multiple comparisons was applied (Benjamini-Hochberg). A change is treated as firm only if it is significant in (b) and (c) after that correction and all three comparisons point in the same direction.
• Measurement changes: some metrics jump abruptly and identically for every team (forced turnovers, ball recovery time, offers to receive, receptions in the final third, among others). A real change in the game rarely moves every team the same way; the most likely explanation is that FIFA changed how these are measured. They were not used in the text.

A.4 Goals

2022 had 172 goals in 5,910 minutes of play; 2026 had 308 goals in 9,630 minutes (including extra time, excluding penalty shoot-outs). That gives 2.62 and 2.88 goals per 90 minutes.

A.5 Low block, pressing and contested play

• Phases of play (FIFA): the report gives the share of each team’s out-of-possession time spent in each phase (low, mid or high block; low, mid or high press; counter-press; recovery). Phases can overlap, so they do not add up to 100%.
• Low block: the share of out-of-possession time spent in a low block went from 19.1% to 23.2% (+21% raw; +27% for the same 26 teams). It is not explained by the scoreline: controlling for the minutes each team spent winning or losing, the increase stays the same (+23% vs +23%). Leading does push teams into a low block, but it does not explain the difference between tournaments.
• Shape of the low block: its length (last to first outfield player) dropped from 25.2 to 22.5 m. The back line stayed in place, between 17.8 and 18.2 m from goal (the small rise is not firm). So the block got shorter because the most advanced player defended about 2 m deeper. Its width did not change (35.7 vs 35.4 m).
• Pressing: pressures fell 8% and their average duration rose 6% (1.48 to 1.58 s). Total time spent pressing (count × duration) did not change: −2%, interval −8% to +3%. Ball recoveries per 100 pressures did not change either (+3%, not significant). The share of pressures steering the opponent inside fell 10% (23.1% to 21.1%), meaning pressure pushed opponents more towards the touchlines.
• Contested play and transitions: time with the ball contested fell 20%, aerial duels won 18% and physical duels won 45% (the latter with few cases per match). Ball progressions (carries) rose 20% and time in counter-attacks fell 22%. All of these are firm and are not explained by the scoreline.
• Unopposed build-up: the share of in-possession time spent building up without opposition rose about 9% (33.5% to 36.1%). It is significant when adjusting for team strength and borderline for the same 26 teams (q = 0.08), so it should be read as supporting evidence rather than a firm result.

A.6 Physical data, heat and substitutions

• Speed zones: player distances in each speed zone (above 25 km/h; high-speed runs), which the report defines the same way in both tournaments.
• Effect of heat: measured with each player compared with himself. For every 10 °C more in the stadium, a player covers about 270 m less per 90 minutes (−2.7%), makes 9% fewer high-speed runs and 7% fewer sprints. For pressing, heat explains the drop in high pressing (which disappears once heat is controlled for) and part of the drop in the number of pressures. It does not explain the rise in low-block time or the drop in contested play.
• High intensity, 2026 vs 2022: controlling for temperature and team strength, distance above 25 km/h fell 9% and high-speed runs 5%, with the same total distance. FIFA does not report whether the calibration of its tracking system changed between tournaments, so part of this could be measurement.
• No wear through the tournament: comparing each player with himself (3,088 appearances of 60 minutes or more), intensity does not fall from one match to the next in either World Cup: the change is ±0.5% per match and not significant. The same holds using accumulated minutes or days of rest.
• Substitutions: in the first phase (group stage, plus the round of 32 in 2026), matches in which a team used all 5 substitutions went from 59% to 81% (interval +10 to +33 percentage points). No difference in the last 16.
• Concentration of minutes: share of minutes played by each team’s 11 most-used players, excluding the third group match, for teams that played 5 or more matches: 82.1% vs 83.4% (not significant). Quarter-finalists only: 82.1% vs 81.7%.

A.7 Goals from outside the box

Unit: shots excluding penalties; for conversion and goalkeeping, shots on target. "Per team per 90 minutes" means one team in one match.

In the text, goalkeeper "effectiveness" refers to the share of on-target shots the goalkeeper saved and held (46% → 34% from outside the box, unchanged at about 19% from inside).

Placement and threshold. Conversion of on-target shots from outside the box by the distance between the point where the shot crossed the goal line and the nearest post or crossbar:

With the 0.5 m threshold the difference is significant (Fisher, p = 0.008). With 0.3 m, 0.75 m or 1 m it is not, because the samples are small. The increase is concentrated between 0.5 and 1 m from the post; shots more than 1 m from the post are rarely converted in either year. Controlling for placement, the 2026 effect barely changes (odds ratio 1.9; p = 0.053).

• Shooters: the 38 goals of 2026 were scored by 33 players from 21 national teams; nobody scored more than 2. Excluding those who scored 2, conversion is still 13.1% (vs 8.7%).
• Goalkeepers: the goalkeeper is the one with most minutes for the defending team. In 2026, 27 of 55 goalkeepers conceded goals from outside the box; Alisson conceded the most (3); Courtois, Pickford and Bounou conceded 2. By the ranking of the defending team, conversion went from 7.0% to 17.9% against the top 16, from 4.7% to 13.6% against teams ranked 17 to 32, and stayed at 16.7% (vs 20.8%) against teams ranked 33 or lower. New teams conceded 14.3%, the same as the rest: this is not an effect of the expansion.
• Consistency: the effect holds in the last 16 (4.4% vs 19.5%), so it is not a lack of adaptation early in the tournament.
• The ball: a wind-tunnel study (Asai et al., Scientific Reports, July 2026) compared the Trionda with the Al Rihla. At 30 m/s without spin, the fluctuation of the lateral force was 1.31 vs 0.96 N (36% higher) and the lift fluctuation was also higher, with very large effect sizes. The authors link this to more irregular flight with little spin, although the test did not measure real trajectories.
• Quotes: Joe Hart’s comments were reported by Goal.com and Flashscore during the 2026 group stage. The Casillas and Buffon quotes about the Jabulani (2010) are as reported by Flashscore.

A.8 Gap between teams

• Heavy wins: in the first phase (2022: 48 group matches; 2026: 88 group and round-of-32 matches), matches won by 3 or more goals went from 10% to 23% (p between 0.05 and 0.08 depending on the test; group stage only, 10% vs 25%, p = 0.05).
• Dominance on the pitch: by match, the dominant team’s share of shots (66% vs 66%) and of possession (61% vs 61%) did not change. The relative gap in strength, measured with the log of the ranking, did not change either (1.27 vs 1.29).

A.9 Did teams change their play because of the format? (basis for the "Theory")

If coaching staffs had changed their style to save energy for a longer tournament, the change should be larger in the first phase or among the strongest teams. Neither shows up:

None of the differences between groups is significant, and the scoreline does not explain the changes either (see A.5). The only difference by phase: the last 16 of 2026 saw more long-ball play (+30%).

A.10 Injuries and early substitutions

• Early substitution: taken off before the 60th minute, outside half-time. The report does not give the reason. In the last 16 of each tournament every individual substitution was checked against press reports: 2 exits for physical reasons in 2022 (both by the same player) vs 8 in 2026. The sample is small and no statistical test was run.
• 2026 final: Argentina replaced three defenders for physical reasons: Lisandro Martínez (44’, thigh injury), Gonzalo Montiel (57’, reported discomfort, not confirmed) and Cristian Romero (70’, injury).
• Caution: not every injury is caused by wear (there are knocks and fractures), and there is no official record of the reason for each substitution.

A.11 Limitations

• FIFA’s report is the only source of match data and was not checked against an independent source.
• Some metrics probably changed definition or calibration between tournaments (A.3); they are not used for conclusions.
• The final stages have 16 matches per tournament: only large differences can be detected.
• "End-of-season fatigue" and "a general change in style" are explanations consistent with the data but not verified: they would need club data.
• In the later stages only the teams and players who kept going remain (survivorship bias).

A.12 References

• FIFA. Enhanced Football Intelligence – Explanation Document, v1.0 (2022) and v1.1 (2023). FIFA Training Centre.
• FIFA. Post-Match Summary Reports, FIFA World Cup Qatar 2022 and FIFA World Cup 2026.
• Asai, T., Liu, R., Kimachi, K. and Hong, S. (2026). Comparative study of the aerodynamic characteristics and flight behavior of the FIFA World Cup 2026 ball Trionda. Scientific Reports, 16, 31982.
• Flashscore (2026). "Something’s up": Is the 2026 World Cup ball causing problems for goalkeepers?
• Goal.com (2026). Joe Hart questions the Trionda ball.
• Open-Meteo. Historical Weather API.

← Back to the article