Mispriced

Refresh data (about 6 min): every FPL endpoint, every player's match-by-match history, bookmaker odds, then rescore, tiers, optimal squad, value index, reports. Use after the last match of a round.
Quick refresh (about 40 s): prices, ownership, transfers, injury news and availability only. Rolling form stays as it was. Use on deadline morning.
Refresh + refit (about 7 min): full refresh plus a refit of the historical model, rewriting the weights and findings. Every few gameweeks.
Nothing updates on its own: the page shows the last refresh (time in the header) until you press a button or install the scheduled task (scripts\schedule_refresh.ps1: Tue 09:00, Fri 18:00).

Prices change once a day around 01:30-02:30 UK. Ownership and transfers move all day.
Injury news / chance of playing is set by FPL as news breaks, mostly after Thu-Fri press conferences.
Match stats and points update every minute or two during games; bonus is provisional until full time. Unfinished fixtures are ignored by the model, so a mid-match refresh is safe.
Your picks for the coming gameweek only appear after the deadline; until then the squad shown is your last confirmed one (the table says which GW).
Bookmaker odds for the next round appear midweek (Tue-Wed). Before that the coming GW uses the calibrated rolling model.

Tuesday morning: Refresh data. Form, odds and prices have all moved.
Deadline morning: Quick refresh for the latest news and prices.
Every few gameweeks: Refresh + refit so this season's matches enter the model.
Numbers: xPts next is the coming GW (blended with FPL's own forecast); xPts horizon is the next six fixtures from the model alone; transfer gain is the change in best-XI points over the horizon, net of hits.

Your FPL team ID is the number in the URL of your Points page on the official Fantasy Premier League site.
Tier board bands (S/A/B/C/D) are computed live from the players currently shown, ranked by the chosen metric: S = top 6%, A = next to 20%, B = to 45%, C = to 75%, D = the rest. Players with 0% chance of playing or under 45 expected minutes are F regardless of rank.
Value index: 100 = priced in line with the market line, a robust fit of xPts on price across players expected to play with one slope for the whole game (the going rate for a point) and a starting level per position, scored on a version of the model that does not see price or ownership. Above 125 = under-valued, below 75 = over-valued. Fair price is where the output sits on that line; it can sit above any price currently on sale in the position, or below the cheapest. Mispricing is the gap in millions. Captaincy is not in the index.
Press / anywhere to jump to the player search.

Mispricing