Every number on FPL360 comes from a named source, and every recommendation can be traced back to a calculation you can check yourself. This page explains where our data comes from, how often it refreshes, and exactly how each tool arrives at its output. If something here doesn’t match what you see on the site, tell us — that’s a bug, not a feature.
Data sources
- Official Fantasy Premier League API — all player prices, points, form, ownership, fixtures, difficulty ratings, live gameweek scoring, and mini-league standings. This is the same data the official FPL site uses. We do not scrape third-party stats sites.
- football-data.org — fixtures and results for our World Cup 2026 section only.
- Our editors — every article is planned around this data and reviewed before publication. See our editorial policy for how articles are produced and corrected.
How fresh is the data?
- Live match scoring (live table, dashboard): refreshed roughly every 25–30 seconds during matches, with a visible “updated X seconds ago” stamp.
- Player statistics, prices and fixtures (players, planner, fixture difficulty, price changes): cached for 10–60 minutes. Price changes are applied by FPL overnight (UK time), so the price tools are most useful in the evening.
- Assistant projections: recalculated from fresh data each time you load your plan; the written explanation is cached for up to 3 hours.
Definitions we use everywhere
- Form — FPL’s own figure: average points per match over the last 30 days.
- Points per game (PPG) — season total points divided by matches with minutes played.
- Value — total points divided by current price in £m. A 100-point player at £5.0m (20.0) beats a 150-point player at £14.0m (10.7) on value, though not on raw points. Neither number is “better” — they answer different questions.
- Fixture Difficulty Rating (FDR) — FPL’s official 1–5 opponent difficulty. We display it and colour-code it; we do not invent our own ratings.
- Ownership — the percentage of all FPL managers holding a player, from the FPL API. A “differential” on this site means under 10% owned.
How the Assistant’s projection works
The Assistant ranks your squad using one transparent formula:
projected points = base rate × availability × fixture multiplier
- Base rate — a blend of the player’s form, points per game, and FPL’s own expected-points figure (ep_next). Before a season starts, when none of those exist yet, we fall back to a price-implied expectation (more expensive players are priced by FPL to score more).
- Availability — from the player’s status flag and FPL’s “chance of playing next round” percentage. A doubtful player’s projection is scaled down accordingly.
- Fixture multiplier — from the opponent’s FDR and home/away, with FDR 3 as neutral. Easier fixtures scale the projection up, harder ones down.
Captaincy suggestions additionally weight a ceiling factor, because the captain doubles your points and explosive attackers have higher single-match ceilings than steady defenders. The AI layer in the Assistant only explains these numbers in plain English — it never invents statistics, and every figure in its write-up comes from the calculation above.
Projections, upside and Player Explorer metrics
What the numbers are. Points, goals, assists, minutes, starts, price, ownership, status and news shown across FPL360 are taken directly from the official Fantasy Premier League API and refreshed every 30 minutes. 3-GW projection and Upside are FPL360 model estimates and are always labelled as such.
How the projection is built (model fpla_v1). For each player we compute a base points rate from official fields: 55% form, 30% points per game and 15% FPL’s own expected points for the next round. While fewer than three gameweeks are complete this rate is blended towards a price-implied expectation (one gameweek is a tiny sample, so a single big haul is not treated as a points-per-game level). The rate is then multiplied by availability (0 for injured, suspended or departed players; the official chance-of-playing percentage for doubts), by a minutes confidence factor derived from starts, and by a fixture multiplier from the official Fixture Difficulty Rating (FDR 3 neutral; easier fixtures up to about +18%, harder down to about −18%) for each of the next three gameweeks. The three gameweek estimates are summed. We display projections to the nearest point: the model does not justify decimals.
Upside (next gameweek). The next-gameweek projection with a position-based return bonus (goalkeepers +10%, defenders +20%, midfielders +55%, forwards +65%). It is a heuristic “if he returns” figure for captaincy thinking — not a statistical percentile and not a prediction of a ceiling score.
Value (points per £m). Total points divided by current price. Headline value rankings apply a minutes guard (45+ minutes early in the season, 90+ from gameweek 3, 180+ later) so a five-minute cameo cannot top the table; the raw Player Explorer still lists every player.
Form and sample size. “Last 3 / last 5” windows only appear when enough gameweeks are complete for every player being compared, using exactly the same window for each. Early in the season labels say “1-GW sample” rather than implying a trend. Availability categories (injury, doubt, suspension, departed) come from one central classifier applied to the official status, chance and news fields.
What it is not. Projections are descriptive estimates from public data; they are not calibrated probabilities and not advice. We store every gameweek’s projections and compare them with actual points as the season progresses, and will publish error figures (mean absolute error and bias by position and price band) once there is enough data to be meaningful.
What we deliberately don’t do
- We don’t run a proprietary xG or machine-learning prediction model, and we won’t pretend arithmetic is one.
- We don’t guarantee outcomes. A projection is a point estimate; football is high-variance, and a 4-point projection losing to a 2-point projection on any given weekend is normal.
- We don’t publish player statistics we can’t source from the official API.
- We don’t accept payment to feature or recommend specific players.
Found an error?
If a number on any tool disagrees with the official FPL site, the official site wins — our copy of the data may be up to an hour old, or we may have a bug. Report it via the contact page and we’ll correct it. Corrections to articles are handled under our editorial policy.