Methodology · engine v7

How the engine thinks.

Five principles, one engine. Calibrated, transparently graded, and held to the receipt on the calibration page — the recipe is proprietary, the score is not.
we say → players score
4.2 → 4.3
FPL says 4.1, they score 3.2
gameweeks closer than FPL
120 of 120
every player graded, blind test
of scores inside our 80% range
80%
said 80%, happened 80%
Engine
v7
Principle 1

Many signals, not a hot take

— Every projection is built from several independent signals, never from one.

A single signal — last-3-game form, last-season points, captain ownership — is always wrong somewhere. Onside reads several independent views of the same player and combines them, so a recency spike never carries a decision on its own and a slow first half of the season never blocks a player who's just turned a corner.

Nothing goes into the engine on faith. Every part of it is fitted on past seasons and then has to prove itself on a season it has never seen. What does not make the picks better is taken out, however clever it looked.

ReceiptOpen any player on the dashboard: the breakdown card shows where his projected points come from.
Principle 2

Calibrated, not optimistic

— A 70%-haul call has to actually haul 70% of the time.

It's easy to build a model that's confident. It's harder to build one whose confidence is honest. We hold ourselves to calibration — when the engine says a captain pick has a 60% chance to outscore the median, that has to bear out on the population of 60%-confidence picks across the season.

If the calibration drifts, the model is wrong even when its rankings look fine. That's why we monitor it weekly rather than waiting for the season-end review.

ReceiptCalibration is graded per gameweek and exposed on /calibration as the rolling residual + per-position MAE.
Principle 3

Minutes matter more than form

— A 78-minute player is a different beast from a 58-minute player.

The single biggest determinant of whether a player returns is whether they're on the pitch. A rotation-risk forward with great underlying numbers will lose to a slightly worse forward who plays 90, every week, for a full season. The engine treats the minutes question as a first-class gate, not an afterthought — and drops players out of captain-eligibility when the risk is real.

ReceiptMinutes risk surfaces in every player drawer as a rotation indicator. When it crosses the threshold the captain shortlist removes the player automatically.
Principle 4

Every prediction goes on the board

— We don't cherry-pick the wins.

Two records, both public. The blind test: In a blind test of 90,150 predictions over four seasons, players Onside projected at 4.2 points averaged 4.3, and players it projected at 1.2 averaged 1.2. Players FPL's own expected points put at 4.1 averaged 3.2. Onside was closer to the real scores than FPL in 120 of 120 gameweeks. And the live one: every projection this season is frozen before the deadline, graded against what happened, and shown on /accuracy and /calibration. That is the receipt we ask sceptical users to check before they trust anything else on the site.

We publish the misses too. A model that grades itself is the trust signal that matters; everything else is marketing.

ReceiptLive receipt: /accuracy and /calibration, updated every gameweek.
Principle 5

Why we don't publish the exact recipe

— Receipts public, internals private — that's the deal.

Two reasons. First, the work that went into the engine is the work. We tell you what we hold ourselves to and let you check the outputs every gameweek; the precise weighting recipe is the proprietary part.

Second, publishing the recipe is an invitation for any well-funded competitor to clone us in a weekend. We'd rather show our scores than our parameters.

ReceiptIf you're a researcher and need access for verification, email [email protected].
The record

Every prediction — logged.

We don't hide the misses. The calibration dashboard shows every gameweek's residual, the rolling MAE, and the per-position breakdown.