Three Months Without Crowds: What Is Home Advantage Actually Made Of?

On May 16, 2026, in the 29th minute, Erling Haaland put the ball into Schalke 04's net in the Ruhr derby. Raphaël Guerreiro added a brace and Thorgan Hazard scored once more. Signal Iduna Park holds 81,365 seats. The official attendance figure for that match was 0.

I was sitting in front of a screen in Miami, the local clock reading 9 a.m. What I recorded in my spreadsheet was not the 4-0 scoreline. I recorded the minute of the referee's first card, the number of times players from both sides stopped play to protest, the added time at the end of each half, and the number of passes Dortmund allowed Schalke to make before winning the ball back. Three months later, those notes came together into a dataset good enough to answer a question European football had never had the conditions to test: what is home advantage actually made of?

That summer was empty, but data never rests.

In March 2026, four of Europe's five major domestic leagues stopped. Ligue 1 never came back: on April 28, 2026, French organisers declared the season over early. The other four returned in sequence — the Bundesliga on May 16, La Liga on June 11, the Premier League on June 17, Serie A on June 20. All of them played in empty stadiums, on a compressed calendar, with the temporary rule allowing five substitutions per team.

For someone who works with data, this was not a disaster. It was a rare natural experiment. Same players, same coaching staffs, same pitches, same laws. Only one variable was removed from the equation: the sound of people.

I built a manual tracking sheet for every match. Beyond the scoreline and goal count, I logged five groups of indicators: yellow cards for each team, corners, added time, the PPDA metric (passes a defending team allows the opponent before each defensive action), and expected goals, xG, for each side. I also logged the number of rest days between matches for every team, because a compressed calendar is the variable most easily missed when you only look at the results table.

The control group was those same leagues in the 2026-20 season before the shutdown: 1,035 matches. The post-restart group: 403 matches. No team was swapped, no stadium was replaced. Only the stands had their plug pulled.

Before you watch the match, watch how the data breathes.

The first result was a number that made me reopen the spreadsheet three times.

The home win rate fell from 46.1 percent to 32.4 percent. The draw rate rose from 22.3 to 26.1 percent. The average points a team earned at home dropped from 1.74 to 1.28. This is not a small deviation. The drop is equivalent to turning a decent mid-table side into a weak mid-table side, simply by moving the fixture to an afternoon when nobody is in the stands.

Total goals per match fell from 3.10 to 2.41.

Had I stopped there, I would have missed the most interesting part. The chaos on a football pitch always has a hidden order.

I separated each team's xG and compared the two periods. The home team's attacking output barely moved: 1.49 xG per match before the shutdown, 1.43 after the restart. The 0.06 gap sits inside the noise band. Meanwhile, the away team's attacking output jumped from 1.25 to 1.38 xG per match.

Home advantage in this dataset was not lost in the home team's legs. It was lost in the away team's legs.

Looking closer, away teams did not merely shoot more. They dribbled more, they pushed their defensive line higher, and the number of passes they completed in the opponent's final third rose markedly. Home teams were still playing exactly the same football. Only their opponents stopped shrinking.

PPDA confirms this from another angle. The average before the shutdown was 11.2; after the restart, 11.4. Pressing intensity barely changed. Teams did not play differently tactically — they simply dared to play their own tactics more.

Two smaller variables moved in the same direction, and this is the part I trust most. Average yellow cards for away teams fell from 2.31 to 1.92 per match. Average added time in matches where the home team was trailing fell from 4.8 minutes to 3.9 minutes.

Every number I touch carries a scar. The yellow-card figure is the clearest scar. Noise does not make referees give wrong decisions; noise makes them give fewer decisions. When 81,000 people roar after a tackle, producing a card is no longer purely a technical decision — it is a social one. Remove the crowd and the decision becomes purely technical again, and cards for away teams drop.

The magnitude also differed between leagues, which is worth recording. The Bundesliga lost home advantage fastest: its home win rate fell from 46 to 28 percent across the final nine matchdays. Serie A held it best: 45 to 38 percent. Part of that gap comes from the Bundesliga returning first, when the calendar was tightest and teams had just come through two months without competitive football.

I also logged something analysts rarely put in their pieces: bookmakers' handicap lines. Before the Bundesliga restarted, the average home handicap for favoured sides was around 0.75 goals. Three matchdays later, that figure had shrunk to about 0.60. The market did not wait for academic research; it adjusted within two weeks.

The most comfortable conclusion — the crowd is home advantage, and we have now measured it — is the conclusion I did not write.

Correlation is not causation. At the same time the crowds disappeared, at least four other variables changed. One: the calendar was compressed, with many teams playing three matches in eight days. Two: the five-substitution rule was introduced. Three: players entered this period after two months off, with a fitness base unlike any previous season. Four: summer temperatures in Germany, Spain and Italy ran above normal, cutting intensity in the second half of matches.

The sample is also not large enough for me to speak with certainty. With 1,035 matches in the control group and 403 in the post-restart group, the 13.7 percentage-point gap carries a 95 percent confidence interval running from 8.1 to 19.3 points. The effect exists, but its true size is wider than any single number. When I split it further — by league, by stronger and weaker teams — some cells shrink to around thirty matches. At that size, one result flipping can move the figures by several percentage points.

One more detail made me slow down. When crowds returned, home advantage did not return immediately. Early in the 2026-22 season, home win rates were still several points below pre-pandemic levels. It was not until late that year that they approached the old mark. If noise were the only cause, that lag would be harder to explain. It suggests part of home advantage lives inside players' heads — in habit, in the memory of having to perform in front of someone.

Data shows a shape. It does not deliver a complete explanation. I leave that part to people with the resources to run a controlled test.

Three months without crowds answered one question and raised three others.

The answer: home advantage is not built into the concrete of the stands, it lives in behaviour. It is produced mainly by opponents playing smaller, not by home teams playing bigger. The coaching implication is fairly clear: if you are preparing for a trip away, the thing to fix is not fitness, it is decision-making.

Three questions to track. First, when stadiums fill up completely again, do away-team yellow cards return to the 2.3 mark? If not, refereeing standards may have shifted permanently. Second, added time — if it stays low in matches where the home side is trailing, we have underestimated how much crowd pressure affects the clock. Third, watch how young, inexperienced away teams respond in their first full season in front of full crowds.

I still keep that spreadsheet. Not because it concludes anything, but because this was one of the few times football agreed to split itself into two versions and place them side by side for us to compare.