xgEdge

Fantasy Premier League

Gameweek 5

What was written down before this round's deadline, against what happened. Nothing here is re-projected: a round already played cannot be forecast from totals that include its own result.

This is the best fifteen money can buy. For advice about the squad you actually own →

Deadline
19:30 closed

Friday, 18/09/2026

Projected · actual
226 250

2.45 a player, on average

Fixture difficulty

Each club’s chance of winning this round’s fixture less its chance of losing it, in five bands.

  • 1 Win likelier than loss by 50+ points
  • 2 By 20 to 50 points
  • 3 Within 20 either way
  • 4 Loss likelier by 20 to 50
  • 5 Loss likelier by 50+

What the numbers mean

Worked through on this round's most one-sided fixture: the model gives Manchester City 55% to beat Sunderland and 18% to lose, so the win is 37 percentage points the likelier — band 2. The other half of a fixture is the same subtraction the other way round, which puts Sunderland on band 4. Hover any badge to see the figures behind it.

Projected against actual

60 players projected before the deadline, against what the round actually paid.

Where the numbers come from

The model, not the feed

Clean-sheet and goals-conceded points come from this site's own model: a pair of expected-goal rates fitted for each fixture, run through a Dixon-Coles score matrix, so a clean sheet is priced as the probability of one rather than as an average. Goals and assists use Fantasy's own per-90 rates, rescaled to the fixture and shrunk towards the position average in proportion to how few minutes a player has behind him.

The largest source of error

Nobody publishes team sheets in advance. The start probability is estimated from recent rounds and the game's own fitness flags, and it is the largest source of error on this page — which is why it is shown beside every player rather than folded into his total.

The roughest term

Bonus is the roughest term here. It depends on finishing in the top three of a match's bonus-point ranking, which is a position among twenty-two players rather than a rate, so it is estimated from a player's own history and nothing more.