Predicted Lineups Explained: How They Are Made & How Accurate They Are

A predicted line-up is an answer to one question: who is actually going to start? Every fantasy decision you make in the last 48 hours before a deadline depends on it. A captain who is rested, a midfielder who turns out to be a bench option, a defender rotated out of a home game against the bottom club — none of those are analysis failures. They are team-news failures, and across a season they cost more points than almost anything else you can get wrong.

This guide explains what a predicted line-up actually is, how ours is built, how accurate the inputs are when you measure them against the eleven names that really walked out, and how to read a starting probability without either ignoring it or trusting it too much.

Predicted line-ups vs expected line-ups vs confirmed line-ups

Three phrases get used interchangeably and they are not the same thing.

• Predicted line-up — also called an expected line-up; in practice the two are synonyms — is a forecast. Nobody outside the club knows the team yet. It is assembled from selection patterns, press-conference language, injury and suspension records, rotation history around midweek fixtures, and the judgement of people who cover that club closely.
• Confirmed line-up is the official team sheet, released roughly an hour before kick-off. It is not a prediction at all. For most fantasy games it also arrives after the deadline, which is precisely why the prediction has value.
• Rumoured line-up is a single source's guess, usually unsourced and never checked afterwards. It is the thing a predicted line-up should not be.

The distinction matters because it tells you what to do with the number. A confirmed XI is a fact you act on. A predicted XI is a probability you weigh — which is why we publish a start probability per player rather than eleven names in a box.

How a predicted line-up is built

The naive version of this is to pick one well-regarded source and copy it. That works until the week it does not, and it gives you no way of knowing which weeks those are. What we do instead has four steps.

1. Read many independent sources, per league

We read a set of line-up sources for every competition we cover — national outlets that cover one league properly, dedicated fantasy analysts, data providers, and specialist accounts who follow a single club. Some are free, some are paid feeds. The set is different in every league, because expertise is local: the people who call a Bundesliga XI best are not the people who call a Serie A XI best.

2. Score every source against what actually happened

This is the part almost nobody does, and it is the whole reason the rest works. Once a match kicks off, the real eleven is known. So every prediction that source made becomes checkable: eleven names, each one either right or wrong.

We keep that score on a rolling window. In the most recent scoring run, covering 11 June to 7 September 2026, that is 1,886 scored line-ups and 20,600 individual player calls across 25 sources. Across all of them, 85.8% of predicted starters actually started. In the Premier League, where seven separate sources are scored against each other, the best of them run above 91%.

Two things fall straight out of that measurement, and neither is obvious from the outside:

• No source is good everywhere. The same outlet that calls one league above 91% calls another in the mid-eighties. A single-source approach inherits that variation silently.
• An 85% hit rate is not an 85% chance of a correct XI. Eleven calls at 85% each, if they were independent, would give you a fully correct eleven roughly one time in six. They are not independent — the errors cluster, because the players a source gets wrong are the rotation decisions everyone gets wrong — but it explains why a predicted XI with one or two names off is the normal case rather than a failure.

3. Weight the sources by that score — and de-duplicate them

A measured accuracy becomes a weight, so a source with a better record in that specific league counts for more there. Two refinements stop that from going wrong:

• Shrinkage. A source with five scored line-ups has not earned the same confidence as one with three hundred, so a thin record is pulled back toward the average rather than taken at face value. And because eleven names inside one XI are correlated, we count line-ups, not player-predictions, when deciding how much evidence there is.
• Redundancy clustering. A hit rate cannot see copying. Two sources that publish the same XI because one is reading the other are not two opinions, and counting them twice puts a thumb on the scale for whatever they both copied. Sources that agree at near-identical rates are grouped and share a single vote. This is the piece that makes it safe to let the measurement drive the weighting at all.

4. Output a probability, not a team sheet

The blend produces, for every player, a chance of starting — and that feeds projected minutes, which in turn feeds the predicted points you see everywhere else on the site. A player at 95% and a player at 60% are both "in the predicted XI" if you flatten it to eleven names, and they are completely different propositions for a captaincy decision.

When the confirmed team sheet lands, the prediction is replaced by the fact, and every projection built on it moves with it.

How to read a predicted line-up properly

• Read the percentage, not the position in the list. The useful question is never "is he in the XI" — it is "how sure are we". Above about 90% is as close to nailed as a forecast gets. Between 50% and 75% is a genuine rotation risk and should change what you do with the armband.
• A low probability is information, not noise. A player at 55% who is priced and owned as a nailed starter is one of the clearest edges available in fantasy, because most of your mini-league is treating him as 100%.
• Expect it to move. Predicted line-ups get more accurate closer to kick-off, as press conferences and late fitness news land. A Thursday prediction and a Saturday prediction are not the same prediction, and the Saturday one is better.
• Check it against your own deadline. In games with per-match deadlines, the confirmed XI for an early kick-off is public before you pick for the late one. In games with a single weekly deadline you never get that, and the prediction is all you have.
• Do not treat one wrong name as a broken model. A single rotation call going against you tells you nothing about whether the probability was right. Over a season, backing higher start probabilities beats backing reputations — and that is the only claim a forecast can honestly make.

Why predicted line-ups are worth more than they look

Most fantasy content argues about which good player to own. Predicted line-ups answer a question one level below that: will the player you already own be on the pitch at all. A zero from a benched premium is worse than a poor return from a starter, because the poor return was priced in and the zero was not. It is the cheapest points leak in the game to close, and closing it requires no opinion about football whatsoever — only an accurate answer to a factual question about the next 48 hours.

That is also why it is the one part of fantasy analysis improved by measurement rather than by argument. There is a right answer every week, it becomes public an hour before kick-off, and anyone who wants to can go back and check.

Where to see them

• FPL predicted line-ups — projected starting XIs for every Premier League club, with a start probability and projected minutes per player.
• UCL Fantasy predicted line-ups — Champions League league-phase XIs, where rotation around domestic fixtures matters most.
• FPL player picks — every player ranked by predicted points, which already has the start probability priced into it.
• FPL team planner — model a transfer or a captaincy switch before you commit it.
• What predicted points (pP) are — how projected minutes become a points projection.

Frequently asked questions

How accurate are predicted line-ups?

Measured over 1,886 scored line-ups between 11 June and 7 September 2026, 85.8% of predicted starters across all our sources actually started, with the strongest Premier League sources above 91%. Accuracy rises as kick-off approaches and official team news lands.

What is the difference between predicted and expected line-ups?

Nothing — they are two names for the same thing: a forecast of the starting eleven before it is announced. Both are different from a confirmed line-up, which is the official team sheet released about an hour before kick-off.

When are predicted line-ups updated?

Continuously through the build-up to each round, as team news, press conferences and injury updates arrive — and again when the confirmed XI is released, at which point the prediction is replaced by the actual team.

Why does a player have a start probability instead of just being in or out?

Because "in or out" throws away the part you need. A 95% starter and a 60% starter are both in an eleven-name list and are completely different captaincy propositions. The probability is also what feeds projected minutes, and therefore the predicted points for that player.