How to build a wildcard squad with an optimiser

When to play it

The three good reasons to wildcard are structural: your squad has three or more non-playing assets, a large block of your team is entering a bad fixture run while another block is entering a good one, or you need to restructure before a double gameweek or chip sequence.

The bad reason is emotional — a single poor gameweek. Wildcarding after one bad week converts a variance event into a permanent loss of a chip.

What the optimiser is solving

Maximise total expected points across the planning horizon, subject to: a 100.0 budget (or your current squad value), fifteen players in the mandated position split, no more than three from any one club, and a valid starting eleven formation.

This is a constrained integer problem with tens of millions of feasible squads. A human cannot search it; a solver can, in under a second.

Why you should not blindly copy the output

An optimiser maximises the objective it was given. If the objective is a four-gameweek point sum, it will happily load you with players whose fixtures fall off a cliff in GW+5, and it will pick bench fodder that never plays because non-playing players cost nothing in the objective.

Sensible guardrails: require a minimum minutes probability on every starter, require at least one playing substitute, and cap exposure to any single club's clean sheet beyond the three-player rule if the model is leaning heavily on one defence.

Structure choices

• —Premium-heavy attack with cheap enablers maximises ceiling and rotation risk simultaneously
• —Balanced 5-4-1 style spreads risk and usually wins on floor
• —Double defensive coverage of one elite defence is efficient but correlated — one bad night costs you twice
• —Genuine bench fodder is only acceptable if your first eleven is fully nailed

Comparing your build with the model's

The most useful workflow is not accepting the optimiser's fifteen, it is diffing them. Paste your intended squad, or a pundit's, next to the model's and look only at the disagreements. Each disagreement is either an assumption you hold that the model does not, or a blind spot in your build.

You will usually keep two or three of your own picks on judgement grounds and adopt the rest. That is the correct outcome.

After the wildcard

A wildcard squad should not need a hit for at least three gameweeks. If your build requires an immediate repair, the plan was wrong. Sanity check the following three deadlines before confirming.