How the verdict is graded

Every call is judged on projections frozen before the deadline, against what you would have fielded without it. Hindsight never counts as skill: a captain who blanked can still have been the right call.

Four words

  • Projected: points on the projections frozen before the deadline. The grade is always projected.
  • Actual: the points FPL scored for the same comparison. It is shown beside every grade and never changes it.
  • Luck: actual minus projected. Nothing else is called luck.
  • Hindsight: against the best choice that was open, on actual points. It is a review, never a grade.

The judge

Before each deadline, Deadline Room freezes its projection of the coming gameweek for every player in the game. The data refresh rewrites the freeze once a day, in the early hours, so the verdict grades against the last one written before the deadline, often many hours before it, and nothing later. Every card says when that was and which engine wrote it. The grade is only as good as those projections, and they have run hot near the top of the board this season.

A second freeze holds each player’s points for the next five gameweeks, which grading a transfer needs. It is stored from GW11, so transfers are graded from that deadline on.

A gameweek with no freeze is not graded. It never shows a zero in place of a grade.

Decisions

Each call is graded against your default: what you would have fielded without making it. The projected points of what you chose, minus the projected points of the default. A good call scores above zero, a bad one below.

  • Captain: against the player in your starting XI whom the most managers captained that week. That is FPL’s own most-captained player when he is in your XI. Otherwise it is read from our sample of 600 managers across the table, then from a sample of the top 10,000, and the sentence names which. Ownership only breaks a tie. Under Triple Captain the gap counts twice. A week with no captaincy data is not graded.
  • Starting XI and bench: against last week’s XI and bench order, with each player you sold replaced by the one you bought, and anyone the freeze had down as out swapped the way FPL’s auto-subs would. Your first gameweek is not graded.
  • Transfers: each move’s projected points over the five frozen gameweeks, in minus out, less its share of any hit, against rolling the transfer. A move made for later fixtures is not punished for the week it was made. That total is spread evenly over the five weeks, so the week’s grade adds like with like. Free Hit and Wildcard moves are not graded.
  • Chips: Triple Captain and Bench Boost against holding them. A chip you play now is gone for every week after, so holding is worth the best week you could still have played it in before it expires. For a team with a Deadline Room account, the chip advisor names that week and its points at each deadline, already counting likely double gameweeks and the risk that your players are injured or dropped by then, and the verdict freezes both. Playing on a week the advisor rated below a later one scores below zero. Where nothing was frozen, holding is worth the chip’s best other week in the five frozen gameweeks, which cannot see further ahead, and the grade says so. With neither freeze, the chip is not graded.

Decisions that total within 2 points of the default make a neutral week: about one player’s projection error. The total is judged as shown, to one decimal place.

Beside each call is its actual: what the same comparison scored once the gameweek was over, after FPL’s auto-subs and the armband passing to the vice. Start a forward over last week’s at +3.3 projected, and if he outscores him by 7 the call reads “+3.3 projected, +7 actual”. The grade stays +3.3. A transfer’s actual is only the first of its five gameweeks, and says so. A chip has no actual, because the week it was held for was never played with it.

Separately, the breakdown shows what you could have had: the best choice open to you, less what you chose, projected. It is there for reference and never counts in the grade.

Luck

Your actual points minus the projected points of what you chose. A player already out at the deadline is projected the way FPL covers him: the first bench player who keeps the formation legal, and the vice for a captain. An injury in the warm-up, and every goal nobody saw coming, land here.

Your luck splits in two. The luck on your captain and XI calls is each call’s actual minus its projected value. The rest is the luck the calls did not change. The two add up to your luck.

Luck is measured against the managers around your rank. They had luck too, on the players they fielded, and only the difference moves you. A haul everyone owns moves you nowhere. Within 3 points either way is a neutral week.

Without a sample of those managers, as in the first gameweeks, luck against the field is unknown: the card says “No sample” and the line speaks only to your calls. While a gameweek is in play the points are partial, so luck is not measured until every match is over and its bonus confirmed.

Points now, places later

Your rank change is FPL’s own, so it is shown in places. Decisions and luck are shown in points.

Turning them into places means estimating what the managers around you scored, from a sample of them. We tested that conversion against real rank moves and it was not close enough, so we will not show you an estimate dressed as a fact. The players who cost you are shown the same way: the points the field gained on you through them.

Places for decisions and luck come back once the estimate of the field passes its test.

Over a season: your overrides

One week’s actual is mostly the football. Over a season, outcomes judge one thing: the calls where you overrode the engine. Each is a gameweek where your captain or starting XI differed from the engine’s pick, frozen before the deadline. The record shows how many overrides you made, what they were worth projected, and the actual points they won or lost against the engine’s picks. Calls where you followed the engine are left out, because they say nothing about you.

Below 10 overrides the record says there are too few to judge. That minimum is our policy, not a fitted number. A short run is noise and is never called skill. Transfers are not judged, because no transfer plan is frozen before the deadline. Chips are not judged, because a chip has no actual for a week it was not played in.

The line

The one-line verdict comes from a fixed table, never written by a model: good or poor calls, against a lucky, neutral or unlucky week, or the calls alone when luck is unknown. It never mocks. Honest mode, which you turn on yourself, is a little blunter.

Get your verdict