Home advantage in football: what it's worth and why it exists
Published
Every football fan believes in home advantage, and almost nobody can say what it is. Home advantage in football is the most-cited, least-examined force in the game: pundits invoke it, bettors pay for it in every price, and most tipsters treat it as a vibe rather than a number. It is a number — one of the oldest and most robust effects in sports statistics — but it is smaller than folklore says, it varies more between leagues than most people expect, and it has been quietly shrinking for over a century. Here is what the evidence actually supports, and how a prediction model turns it into probabilities.
What home advantage in football is worth
Across the 16,992 finished matches in our nine-league dataset (2021–2026), home teams won 43.9% of the time, draws took 25.6%, and away teams the remaining 30.4%. That gap — roughly twelve percentage points between winning at home and winning away — is home advantage in its rawest form.
Two things are worth noticing before going further. First, the effect is real and large: no persistent bias of that size exists anywhere else in the match. Second, it is much smaller than intuition suggests. A home team is not “expected to win”; the most common single result of a home fixture is still not a home win. Folklore treats home advantage as a trump card. The data treats it as a thumb on the scale.
Where it actually comes from
Home advantage is one effect with several causes, and researchers have spent decades trying to separate them:
- The crowd. The most popular explanation, and a real one — but not necessarily in the way fans imagine. Much of the measured crowd effect appears to run through the referee: studies consistently find home teams receive fewer cards and more of the marginal decisions, and that these biases weaken when the crowd is removed or distanced from the pitch.
- Travel and routine. Away squads absorb travel, unfamiliar dressing rooms and disrupted preparation. The effect is measurable but modest in domestic leagues, where distances are short; it matters more in continent-sized competitions.
- Familiarity. Pitch dimensions, surface, stadium sightlines. Historically this was a bigger deal — the effect was strongest where grounds were genuinely unusual.
- Psychology and tactics. Home teams simply play differently: more possession, more shots, more territorial aggression. Whether that is a cause of home advantage or a consequence of expecting it is genuinely hard to untangle.
The honest summary of the research: the crowd-and-referee channel carries more of the effect than travel or familiarity, and no single factor explains all of it.
The natural experiment nobody wanted
For decades the causes above were impossible to isolate — every home match had all of them at once. Then the pandemic emptied stadiums, and for the better part of two seasons Europe ran the exact experiment no researcher could ever have commissioned: professional football with travel, familiarity and psychology intact, but no crowd.
The results, replicated across multiple leagues and studies, were striking. Home advantage shrank markedly in the empty-stadium period — in some leagues it nearly disappeared — and refereeing statistics moved with it: the home team’s edge in cards and contested decisions weakened when there was no one in the stands to react. When crowds returned, so did most of the effect. It is as close to a controlled demonstration as observational sports data ever gets, and it shifted the academic consensus firmly toward the crowd-referee channel as the largest single ingredient.
Not one number: leagues differ, and the era matters
A model that applies one universal home boost is already wrong, because home advantage is local. In our own dataset, La Liga shows the strongest home edge of the nine leagues we cover, while Ligue 1 and Serie A sit at the weak end — travelling teams win in France at a rate that would look like a data error in Spain. The league-by-league breakdowns in our league guides make the point repeatedly: the “same” fixture, transplanted between countries, deserves visibly different probabilities.
The era matters just as much. In the earliest decades of league football, home teams won well over half of all matches; today’s figure is in the low forties, and the decline has been steady rather than sudden. Professionalisation explains most of it — standardised pitches, easier travel, video refereeing, squads that prepare for away games as carefully as home ones. Any model fitted on decades-old data would systematically over-back home teams today, which is one reason serious models weight recent seasons far more heavily than distant ones.
How a prediction model prices it
Rating-based models like ours don’t treat home advantage as a slogan; they fit it as a parameter, alongside every team’s attack and defence ratings, and re-estimate it continuously from results. Concretely, the parameter inflates the home side’s expected goals and deflates the away side’s before the Poisson scoreline machinery takes over.
A hypothetical shows how much work that one parameter does. Take two perfectly equal teams who would each expect 1.25 goals on a neutral ground. A Poisson grid prices that match at about 36.5% / 27.0% / 36.5% — symmetric, as it must be. Now let a fitted home-advantage parameter shift the inputs to 1.45 expected goals for the hosts and 1.10 for the visitors. The same grid becomes roughly 45.1% home, 26.2% draw, 28.7% away. One estimated number moved the home win by more than eight percentage points — which is exactly why getting it right, league by league and season by season, matters more than most of the exotic features prediction sites like to advertise.
It also means home advantage is where a model can fail quietly. Fit it too large and every home price is inflated; too small and the model looks like it “hates” home teams. Errors of this kind don’t show up in any single match — they show up as miscalibration across hundreds, which is precisely what a public, append-only record exists to catch. Our track record grades every published probability against what actually happened, so a systematic home-team bias would be visible to anyone who looked.
What to take from all this
Home advantage in football is real, modest, local and shrinking. It is worth a few tenths of a goal, not a guaranteed outcome; it varies enough between leagues that a single universal boost misprices matches; and its largest ingredient seems to be the crowd acting on the referee, not some mystical power of familiar grass. When you read any match probability — ours included — part of what you are reading is one fitted parameter’s opinion about all of the above. The only way to know whether that opinion is any good is to check it against results, in public, over a sample large enough to mean something.