OPTIMIZATION ENGINE v1.0

Beat the field.
Not the projections.

EdgeRank is an AI-powered FPL optimizer that maximizes rank improvement against the field — not raw expected points. Every pick is computed against ownership-adjusted differential EV, environmental friction, and 10,000-path Monte Carlo bracket simulation.

OBJECTIVE

Rank Δ vs. Field

SOLVER

MILP + LP

SIMULATIONS

10,000/cycle

DATA

Opta + StatsBomb

WHY EDGERANK IS DIFFERENT

Rank-optimized objective

Most tools maximize projected points. EdgeRank maximizes the probability of finishing above other managers. Same inputs, completely different optimization target.

Leverage-adjusted EV

EV_i × (1 − O_i)^α. A pick held by 60% of the field contributes structurally less to rank gain than an equivalent pick at 8% ownership. Static alpha is wrong. EdgeRank calibrates dynamically.

Environmental friction model

Altitude. Heat index. Travel fatigue. These are first-class inputs — not afterthoughts. Altitude-induced VO₂ degradation and cumulative transit friction are modeled continuously from tournament data.

Monte Carlo fixture weights

10,000 post-matchday bracket iterations feed live into the LP solver. Fixture longevity weights update in real time. The solver and simulator share a live feedback loop — not a batched one.

SIGNAL ARCHITECTURE

SET PIECE COEFFICIENT

Penalty takers, corner delivery specialists, free kick executors. Particularly relevant for defenders and defensive midfielders whose open-play xA understates actual contribution.

RECENCY-WEIGHTED FORM

Temporal decay function. Recent matches weight higher than tournament averages. Context-sensitivity — a poor result against a top defensive side is treated differently from the same result against a low block.

ENVIRONMENTAL FRICTION

E_f = 1 − δ × f(Alt, HeatIndex). Altitude-induced VO₂ max degradation and heat index multipliers recalibrated post-tournament cycle against predicted versus actual differentials.

TRAVEL FATIGUE VECTOR

Cumulative F_i tracking kilometers, transit modes, and time-zone shifts. Sprint volume and defensive action success metrics discounted against recovery days available.

ROTATION RISK FLAG

When nation probability of securing bracket seed reaches ≥85%, model triggers tiered rest-cycle flag. C_i automatically lowered for elite assets.

LINEUP PROBABILITY COEFFICIENT

C_i ∈ [0, 1] applied to every EV vector. Derived from manager rotation history, injury telemetry, and group advancement math. NLP scoped strictly to injury and fitness signals.

SAMPLE OPTIMIZATION OUTPUT

EdgeRank Solver — GW28, 2025/26

        PLAYER
        EV
        LEVERAGE
        OWN%
        C_i
        ENV FLAGS
      

        Son (H)
        8.42
        0.91
        6.1%
        0.94
        ALT +2.1
        CAPTAIN
      

        Wirtz
        7.18
        0.84
        4.3%
        0.91
        HEAT
        START
      

        Oyarzabal
        6.77
        0.62
        11.8%
        0.88
        —
        START
      

        Comeret (H)
        5.94
        0.58
        2.9%
        0.97
        REST
        DIFF
      

        Saka
        7.03
        0.29
        38.4%
        0.82
        —
        OVER
      

Differential picks (low ownership, high leverage) are marked separately from the core squad. The solver optimizes for rank improvement, not points accumulation.

The field is 11 million managers.
Most of them are using the same model.

EdgeRank exists for the ones who want to be in the top 10,000 — not the top 10,000 who use the same tools as everyone else. The optimization is different. The signals are proprietary. The objective function is rank, not points.

FPL optimization for managers who play to win.

Beat the field.Not the projections.

Beat the field.
Not the projections.