The Model XI: our engine's own fifteen, and why Haaland is not in it
A 3-5-2 projecting 65.92 points with B.Fernandes captained — built on six-week horizon value, market-implied fixture strength, and one deliberate act of heresy.
This is the squad we will actually run all season, transfers and all, and it starts with the decision most managers would call reckless: no Haaland. The engine prices him at 7.8 expected points against Bournemouth and nothing about that number is bad — it is simply what £15.5m buys. Spread across cheaper players, the same money buys more. B.Fernandes at £12.0m projects 8.85 expected points away at promoted Hull, the highest figure on the entire board, and he wears the armband: 17.7 doubled.
The spine under him is unglamorous on purpose. Everton host Crystal Palace, and the model likes the whole block: Tarkowski projects 4.67 with a 45.7% chance of banking the defensive-contribution bonus, and Ndiaye 5.13 at a value of 0.856 points per million — the best of any midfielder this week. Raya (4.63, with 2.2 of it from clean-sheet scenarios), Virgil (4.28) and Senesi (4.35, 58.4% defensive-contribution probability) give the back line three routes to points that do not require anyone to score.
Fifteen players, fifteen projected starters, £100.0m to the penny — no bench fodder, no dead money.
Up front, Thiago (5.89) and Watkins (4.94) carry the forward line at £8.0m each, and the midfield is completed by Gibbs-White (4.83), Szoboszlai (4.71) and Le Fée (4.79 at just 9.9% ownership). The bench is the quiet flex: Sánchez, N.Williams, Calvert-Lewin and Shaw all project to start their matches, which means an automatic substitution costs us real points instead of zero — and keeps a Bench Boost live whenever we choose to play it.
One free transfer a week means you live with your opening squad for a month, so this was never a week-one decision. Every pick was chosen on six-gameweek projections, simulated week by week: gameweek 1 from live market odds, weeks 2 to 6 from fixture-difficulty priors calibrated on those odds and replaced by real prices as each week arrives. Over the full run B.Fernandes projects 42.3 points to Haaland's 36.9 at a price 3.5 lower, Thiago 33.8, Ndiaye 31.3, and the Everton block holds all six weeks. The same lens is why Saka misses out despite the best week-one fixture in the game: 6.1 this week, then 3.4 and 3.5 in the two weeks after.
Why publish a squad that ignores a player three-quarters of the game owns? Because the point of this site is that the numbers are the argument. Owning Haaland is insurance against the crowd; our model is paid to take the expected-value side of that trade. The Consensus XI on this site owns him, captains him, and follows the crowd all the way down — and every week we will publish both scores. One of these philosophies is going to be embarrassed by May. We are genuinely not sure which.
Bottom line: The Model XI projects 65.92 points in gameweek 1 with B.Fernandes captained — the engine's pure expected-value squad, run publicly all season against the consensus alternative.
The data
| # | Player | xPts | Ceiling | Captain EV | Pts/m |
|---|---|---|---|---|---|
| 1 | Raya ARS | 4.63 | 8.17 | 9.25 | 0.77 |
| 2 | Virgil LIV | 4.28 | 10.97 | 8.57 | 0.66 |
| 3 | Senesi TOT | 4.35 | 10.55 | 8.69 | 0.72 |
| 4 | Tarkowski EVEDifferential | 4.67 | 11.16 | 9.33 | 0.78 |
| 5 | B.Fernandes MUN | 8.85 | 19.70 | 17.70 | 0.74 |
| 6 | Ndiaye EVE | 5.13 | 15.04 | 10.27 | 0.86 |
| 7 | Gibbs-White NFO | 4.83 | 13.53 | 9.67 | 0.60 |
| 8 | E.Le Fée SUNDifferential | 4.79 | 13.02 | 9.59 | 0.80 |
| 9 | Szoboszlai LIV | 4.71 | 12.49 | 9.43 | 0.67 |
| 10 | Thiago BRE | 5.89 | 14.16 | 11.78 | 0.74 |
| 11 | Watkins AVL | 4.94 | 13.15 | 9.88 | 0.62 |
| 12 | Sánchez CHEDifferentialBench | 3.14 | 7.75 | 6.29 | 0.63 |
| 13 | N.Williams NFOBench | 4.10 | 10.80 | 8.20 | 0.82 |
| 14 | Calvert-Lewin LEEBench | 3.97 | 11.68 | 7.95 | 0.66 |
| 15 | Shaw MUNBench | 3.98 | 7.99 | 7.95 | 0.88 |
Bottom line
The Model XI projects 65.92 points in gameweek 1 with B.Fernandes captained — the engine's pure expected-value squad, run publicly all season against the consensus alternative.
Get the next round the moment odds drop
50,000 simulations per matchday — captains, EV and match predictions in your inbox before lock.
Double opt-in. No spam, no tracking. Unsubscribe anytime.
How we get these numbers. Market odds (de-vigged) → Dixon-Coles scorelines → 50k Monte-Carlo simulations, scored on the official Fantasy Premier League points table. Ceiling = the average of a player's best 15% of our 50,000 simulations — the score when his best realistic games happen, not a fantasy cap. Every figure here is machine-readable at /api/fpl/gw1/our-squad.json.