FPLHERMES
Log inStart for free
Powered by 500 sims per fixture

Better data.
Smarter moves.
Win your mini-league.

FPLHERMES simulates every fixture 500 times using Monte Carlo AI to deliver powerful predictions, captaincy picks and transfer advice.

Start your edgeSee how it works
FPLHERMESGW6

Predictions

Projection matrix for Gameweek 6 onward.

GW6GW7GW8GW9GW10GW11
PositionTeamPriceMin prob
S

Saka

ARS · MID

9.53
G

Gibbs-White

NFO · MID

7.84
BF

B.Fernandes

MUN · MID

7.18
G

Gabriel

ARS · DEF

7.04
P

Palmer

CHE · MID

6.47

Backtested · 2025/26 walk-forward replay

2,241 points last season, with the chips played on the engine's own timing.

2,241
2025/26 replay, chips timed by the engine
2,172
before chips were played
59.0
average points per gameweek
~2,300
typical top-1% finish, chips included

No lookahead: every pick used only data available before that deadline. Bench Boost, Triple Captain and Free Hit are scored from the replayed season; the wildcard figure is an estimate, and the top-1% line is a population reference, not a published rank.

Accurate predictions

500 simulations per fixture.

Smarter insights

Captain picks, transfers, differentials.

Beat your league

Find the edges your rivals miss.

Save time

All the data. None of the noise.

Trusted by FPL managers · 667 players simulated for gameweek 6

Backtested · full 2025/26 replay

2,241 points last season, chips timed by the engine.

FPLHERMES 2,172 Template 2,076
2,241
with chips, engine-timed
2,172
before chips
+69
from the chip layer
+96
vs template team

Walk-forward, no lookahead. Bench Boost, Triple Captain and Free Hit scored from the replay; the wildcard contribution is an estimate.

$5 a week. Three days to decide.

Sign in with Google, add a card, full engine unlocks. No charge for three days, cancel anytime.

  • ›xPts for every player, every gameweek
  • ›Captaincy floor and ceiling
  • ›4-gameweek transfer planner
  • ›Injury intelligence
  • ›Mini-league + H2H tools
  • ›Weekly accuracy audit

FPLHERMES Pro

$5/week

3-day trial, then billed weekly · cancel anytime

Start 3-day trial

What you get

xPts engine

500 simulations per fixture. Mean, floor, ceiling.

Captaincy

Safe armband by floor, ceiling armband by p90.

Transfer planner

4 gameweeks ahead, hits priced against gain.

Injury intel

Official flags plus reporting, scaled into xPts.

League warfare

Effective ownership, threats, edges, H2H duels.

Self-audit

Weekly accuracy published in the Model Lab.

A week with FPLHERMES

  1. 1Sat–MonGameweek scores itself
  2. 2TueFresh xPts for every player
  3. 3WedInjury sweep + availability
  4. 4Thu4-gameweek transfer plan
  5. 5FriSafe vs ceiling captain
  6. 6DeadlineMini-league warfare

Live ledger · 2026/27

How the model is doing, right now

Before every deadline we archive the projection. After the final whistle we score it. These are the published numbers — no cherry-picking, updated weekly.

1.68
avg pts MAE, season to date
GW5 · 1.74
latest gameweek MAE
26.3
avg minutes MAE
2,329
player-fixtures scored
GameweekPts MAERMSEBiasMin MAEn
GW11.702.81+0.3928.4500
GW21.552.50+0.2926.8469
GW31.642.51+0.2926.1466
GW41.772.74+0.3325.0453
GW51.742.74+0.4025.0441

Backtest ledger · 2025/26

The replay, and exactly how it was constrained

2,241
chip-adjusted, engine-timed
2,172
base replay, no chips
+69
chip layer contribution
2,288
hindsight ceiling (not achievable)
  • ›Walk-forward: projected only on data available before each deadline
  • ›One free transfer a week · no hits · chips added as a separate, labelled layer
  • ›£100.0m budget · 2/5/5/3 squad · max three per club
  • ›Opening squad picked from 2024/25 priors alone

We won't claim a rank percentile the Premier League doesn't publish. The template benchmark is computed from the same public dataset, and the live engine keeps scoring itself weekly in the Model Lab.

Learn the method

Free, in-depth guides to how AI actually predicts Fantasy Premier League points — and how to use the numbers without being misled by them.

All guides