Why Pre-Season Results Tell You So Little About the Season Ahead

Every July, pre-season friendlies produce scorelines, and every July those scorelines get read as evidence. A heavy defeat becomes a crisis, a run of wins becomes momentum. The question worth asking before the domestic season starts is a simple one: does anything measured in a pre-season match actually predict anything about the campaign that follows?

The short answer is that results predict close to nothing, and that this is not primarily a sample-size problem. It is a comparability problem, which is a different and more serious defect.

The Question, Stated Precisely

There are two distinct claims people make about pre-season, and they deserve separate treatment.

The weak claim is that pre-season results are noisy — that seven matches is too few to conclude much. That is true of any seven matches, competitive or not, and it is the ordinary problem of small samples.

The strong claim is that pre-season data is measuring something other than what league data measures. Not a small sample of the same thing, but a sample of a different thing entirely. That is the claim worth examining, because if it holds, then collecting more pre-season matches would not help at all.

What a Pre-Season Match Actually Is

A friendly is a training session with a scoreboard attached, and almost every rule that makes competitive data comparable is relaxed.

• Substitutions are frequently unlimited or governed by bespoke agreement between the clubs, so the eleven that finishes may share nobody with the eleven that started.
• Some fixtures are played in shortened periods, or as multiple short matches within one meeting, which breaks per-90 arithmetic outright.
• Squads are incomplete. Players returning late from international duty, players carrying knocks, and players whose futures are unresolved are absent by design.
• Trialists and academy players appear in numbers they never will again, sometimes without being formally registered.
• Opposition comes from other divisions and other countries, with no consistent standard of quality across a club's fixture list.
• The objective is not to win. Managers are loading minutes into legs, testing combinations, and protecting valuable players — three goals that all cut against maximising the result.

Any one of these would compromise a dataset. Together they mean a pre-season scoreline and a league scoreline are not the same kind of measurement, even though they are written the same way.

What the Underlying Numbers Do and Don't Survive

It is tempting to skip results and look at underlying performance instead, on the reasonable theory that shot quality is harder to fake than a scoreline. Some of this survives; most of it does not.

Expected goals models are calibrated on competitive football. Applied to a match where defensive intensity is deliberately dialled down and the opposition is two divisions below, the model is being asked about a situation it was never fitted to. The chances it counts are real; the probability it assigns to them is borrowed from a context that does not apply.

Possession and passing numbers degrade for a related reason. Pressing schemes are being rehearsed rather than executed at full intensity, so the pressure a team faces in possession bears little relation to what August will bring. A build-up pattern that works untroubled in July can be dismantled in the opening fixture by opponents actually trying to win.

Physical data is the partial exception, and it is the reason clubs value the period at all. Distance covered, sprint counts, and load management are measured for their own sake rather than as proxies for quality, and they answer a question — is this player fit enough to start — that pre-season is genuinely designed to answer.

There is also a collection problem that sits underneath all of this. Competitive matches in major leagues are covered by dense, standardised data operations; friendlies frequently are not. Event coverage of a mid-July fixture between a top-flight club and lower-league opposition is often thinner, less consistently defined, and gathered under different conditions than the same club's league matches. Data platforms including RubiScore log friendlies as a separate fixture type for exactly this reason: mixing them into season aggregates would corrupt the aggregate. When a pre-season number looks surprising, the first question should be whether it was recorded to the same standard as the numbers you are comparing it against.

The Confounders Are Unusually Large

Even setting aside the format, pre-season fixtures are selected for reasons that have nothing to do with sporting preparation.

Tours are scheduled commercially. Clubs travel to markets where they want exposure, which means long-haul flights, unfamiliar time zones, heat and humidity well outside their normal operating range, and sometimes altitude. A squad three days into an intercontinental tour is not a squad you can compare to itself a fortnight later at home.

Pitch quality varies more than it does in league football, especially in multi-purpose stadiums used for friendlies. Crowd context is inverted: a club may play its "home" fixtures in front of an opposition-supporting crowd on another continent.

And the fixture list is not randomly assigned. Managers deliberately schedule progressively harder opponents as the weeks pass. Any apparent trend across a pre-season — improving performances, tightening defence — is partly a scheduling artefact working in the opposite direction to what it appears.

This Summer Is an Extreme Version

The 2026 pre-season is an unusually poor guide even by these standards, because an expanded World Cup has occupied the middle of the summer.

The tournament's 48-team, 104-match format means more players from more countries have been involved for longer, and the deepest runs finish only weeks before domestic leagues start. The practical consequence is that a large share of first-choice players across Europe's top divisions will have had a compressed break, will report late, and will play little or no pre-season football with their clubs.

That has two effects on the data. The pre-season sides fielded are less representative of the sides that will actually start the league season than in a normal year — so the tactical information is thinner. And the early weeks of the league season will themselves be distorted, as internationals are reintegrated on staggered timelines and managers manage fatigue rather than chase early points.

Tournament summers also concentrate transfer activity late. Clubs wait for the international window to finish before finalising business, so squads in July frequently do not resemble squads in September in either direction.

So What Is Worth Watching?

Pre-season is not information-free. It is simply that the useful information is structural rather than statistical.

• Shape and personnel: which formation the manager returns to repeatedly, and which players occupy the new or unfamiliar roles within it.
• Build-up patterns: how the team tries to progress the ball from the back, since these routines are rehearsed in July and rarely abandoned by August.
• Set-piece routines: often introduced in pre-season and visible before opponents have scouted them.
• The minutes hierarchy: who plays with whom, and for how long, in the final fixtures before the season starts. This is the closest thing to a public team sheet.
• Availability: who is missing, who is being managed, and who has not featured at all.

Notice that none of these is a result. All of them are observations about intention, and intention is the one thing a friendly reveals honestly, because there is no reason to disguise it.

The Verdict, With Its Caveat

Pre-season results have essentially no predictive value for league performance, and underlying metrics from friendlies are only marginally better, because both are drawn from a context whose rules, intensity, and personnel differ systematically from competitive football. The failure is one of comparability rather than volume — which means the fix is not more pre-season matches, and no amount of care in the analysis will recover what the format has removed.

The caveat runs in the other direction, and it matters. The absence of predictive value in the results does not mean pre-season is uninformative about the things it is designed to reveal: fitness, structure, and selection. Analysts who dismiss the period entirely miss the tactical preview sitting in plain sight.

The honest position is narrow and useful. Watch pre-season for what a team is trying to build. Ignore whether it won. And extend the same scepticism to the opening weeks of the season proper, when the samples are small for ordinary reasons and, this year in particular, a great many of the relevant players will still be finding their legs. Fixtures, lineups, and match data across competitive and friendly matches are published on rubiscore.com.