Is the FPL price change predictor accurate?

We snapshot FPL's own price-change predictor seven minutes before midnight every night and check it against the prices that came out. Here is what it got right and what it missed, night by night. Data as of 7 September 2026, Gameweek 3. The bubble map has the same numbers live, refreshed every 15 minutes.

Flagged changes
91
over 5 nights
Happened
90
98.9% of the time
Missed
1
flagged but did not move
Unflagged changes
0
of 90 price moves

Night by night

Night ofFlaggedHappenedUnflaggedGot wrong
6 September 202625250—
5 September 202614140—
4 September 202625240Rice
3 September 202615150—
2 September 202612120—

How this is measured

Every night at 23:53 UK, seven minutes before FPL changes prices, we save a snapshot of the game's own predictor for all 550-odd players. The next snapshot has the prices that came out. A player the predictor had at 100% or more is a flagged change; if his price moved that way at midnight, it happened. A price that moved without being flagged is an unflagged change, and that is the failure that costs you money, because it is the one you could not have seen coming.

Nothing here is our estimate. The prediction is FPL's published number and the outcome is the price in the game, both taken from the official API. We are the scoreboard, not a competing model.

Two caveats worth knowing. The counter resets after each change, so a player who rises cannot rise again the same night. And a snapshot at 23:53 cannot catch a transfer surge in the last seven minutes: a player at 96% at 23:53 who crosses at 23:58 shows here as an unflagged change, which makes this measurement slightly harsher on the predictor than it deserves.

Tonight's list is on the price changes page, and the map colours every player by the same number.

FPLBubbles is an independent tool and is not affiliated with the Premier League or Fantasy Premier League. Data from the public FPL API.