59 terms, every number and word the product uses, defined once here and linked to from everywhere else. What it means, why it matters, and what a good one looks like.
The model's best guess at how many points a player scores next gameweek. It starts from his recent scoring average, then nudges it up or down for how hard the fixture is, whether he is fit, and whether his club plays twice or not at all that week.
Why it matters Every pick, swap and line-up call in the app starts from this number.
Good number 4-5 is a solid starter; 8+ is a premium week; 0 means injured, suspended or no fixture.
The player's average points over his last 6 confirmed gameweeks — the starting point before any fixture or fitness adjustment. For a fit player, weeks he did not play are ignored so a one-off ban does not drag his number down; for an injured player the blanks are kept in, because they are his current reality.
Why it matters It is the raw form reading that everything else multiplies.
Good number 5+ per game over the window is strong form for any position.
A 1-to-5 rating of how hard the next opponent is: 1 is the easiest game on the calendar, 5 is a trip to the champions. The model turns it into a multiplier on projected points — an FDR 1 fixture boosts the projection by 30%, an FDR 5 fixture cuts it by 30%.
Why it matters The same player is worth noticeably more against weak opposition.
Good number FDR 1-2 boosts the projection (x1.30 / x1.15); FDR 4-5 cuts it (x0.85 / x0.70); 3 is neutral.
A week where a club plays two matches instead of one, usually because a postponed game was squeezed in. Two games means two chances to score points, so the model multiplies the projection by 1.8 — not quite double, because managers rotate.
Why it matters A mid-table player with two games often out-scores a star with one.
Good number The x1.8 multiplier means an ordinary 4-point player projects at 7+ in a DGW.
A week where a club has no fixture at all — the game was postponed or the club is busy in a cup. No match, no points: the model projects a flat 0 for every player at that club.
Why it matters A blanking star is worth less that week than any starter with a game.
Good number There is no good number — a BGW projection is always 0.
How far back the model looks when it measures form. Different questions use different windows: the last 6 gameweeks for the scoring baseline, the last 4 to spot a hot or cold streak, and the last 8 starts to judge how streaky a player is.
Why it matters Short windows react fast to a purple patch; longer ones filter out one-off flukes.
Good number 6 gameweeks is the default baseline window.
When the model averages past games, last week counts more than last month. Each game back in time is worth 85% of the one after it — like your memory of matches, sharp on recent ones and fading on older ones.
Why it matters It lets the position ratings catch a player whose role just changed.
Good number The decay factor is 0.85 per gameweek back.
A fitness discount on the projection. A fully fit player keeps 100% of his points; a player rated 75% likely to play keeps 75%, and so on down. Injured, suspended or unavailable players are cut to zero, and a doubt with no official percentage is treated as a coin flip at 50%.
Why it matters It stops the app recommending a player who probably will not be on the pitch.
Good number 1.0 is fully fit; 0.75 or below means real doubt; 0 means do not expect him to play.
Goals the player scored himself. Worth the most points of any single action: 10 for a goalkeeper, 6 for a defender, 5 for a midfielder, 4 for a forward.
Why it matters The single biggest points event in the game.
Good number 15+ in a season is elite for a forward; 5+ is strong for a defender.
The final touch before a teammate scores. Worth 3 points regardless of position.
Why it matters Creative players bank steady points without needing to score themselves.
Good number 10+ in a season marks out a genuine creator.
The team conceded no goals while the player was on the pitch. Worth the most to goalkeepers and defenders, a little to midfielders.
Why it matters It is the bread-and-butter score for every defender and keeper.
Good number 12+ clean sheets in a season is a top defence.
Points for the dirty work — tackles, interceptions, clearances and blocks, bundled into one count per match.
Why it matters It rewards ball-winners who rarely score or assist.
Good number A holding midfielder or centre-back hitting the bar most weeks adds steady points.
Goalkeeper saves. Batches of saves in a match earn points, so a busy keeper behind a weak defence still banks them.
Why it matters It makes some keepers at struggling clubs quietly valuable.
Good number 100+ saves over a season means regular points from saves alone.
Total points divided by games actually played (any minutes at all). It levels the field between a player who missed months and one who played every week.
Why it matters Raw totals flatter the ever-present; per-game shows who produces when he plays.
Good number 4+ is a solid starter; 6+ is premium.
Time on the pitch. Playing at all earns appearance points, and minutes are the best predictor of everything else — no minutes, no points.
Why it matters A nailed-on starter beats a more talented bench option almost every time.
Good number 2,500+ minutes over a season means a trusted starter.
A scout composite — a lens for comparing players, NOT the number the recommendation engine acts on. It blends six ingredients into one score: head-to-head value (30%), consistency-adjusted points (25%), form trajectory (20%), position rating (15%), a penalty for rotation risk (10%) and a bonus for denying rivals a scarce player (10%). Since the L1 rescale every ingredient sits on the same roughly 0-to-10 points-like scale before its weight applies, so those percentages mean what they say — earlier the raw 0-100 position rating drowned out everything else.
Why it matters It gives one comparable score per player, but the engine ranks and gates claims on projected points gain, and the scorecard grades that gain — never this composite.
Good number Higher is better as a comparison between players; do not read it as points.
An expected-points figure that cares about match situation, built from a player's last 8 starts. Chasing a deficit, it leans towards his best games (his ceiling); protecting a lead, it leans towards his worst (his floor); in a close match it just takes his average. With fewer than 4 starts on record it falls back to the plain average.
Why it matters When you are behind you want a boom-or-bust player, not a steady 4-pointer.
Good number Read it like points per game: 5+ is strong, and the gap between two players is what counts.
Is this player heating up or cooling down, once you account for who he has been playing? It compares his last 4 games to his season average, checks whether his next 4 fixtures are easier or harder than his last 4, and discounts anyone who has not been starting.
Why it matters It separates genuinely improving players from ones who just had a soft run of fixtures.
Good number Above 1.2 is a buy signal; below 0.8 is a sell signal; 1.0 is steady as she goes.
How many points a player gives you above two bars at once: the average player your next head-to-head opponent fields in that position, and the best free agent you could pick up instead. It is projection minus both of those.
Why it matters In head-to-head you only need to beat one manager this week — this measures exactly that.
Good number Positive means he both wins you the matchup and beats the free-agent pool; negative means neither.
Extra projected points from FPL's defensive contribution rule, which pays 2 points for a busy defensive shift (clearances, blocks, interceptions, tackles). Defenders and midfielders facing attacking sides rack up more of those actions, so the model scales the player's usual hit rate by how much attacking the opponent does.
Why it matters A defender at a mid-table club facing a big attacking side can quietly bank an extra 2 points.
Good number Around +1.0 or more is a meaningful weekly bonus for a defender or defensive midfielder.
The chance a manager leaves this player out. It is built from how many line-up changes his club has averaged over the last 8 gameweeks, and it rises by half again if the club has a midweek cup or European game within about 4 days.
Why it matters The best projection in the world is worth nothing if he starts on the bench.
Good number It runs roughly 0 to 1 — near 0 is nailed on, above 0.5 is a genuine bench worry.
A rating of how good a player is at the things that score points in his specific position — a striker is marked on shots and goals, a defender on clean sheets and defensive work. Built from per-90-minute rates with recent games weighted heaviest.
Why it matters It compares players fairly within a position instead of by raw points alone.
Good number 0 to 100, higher is better within a position; inside the edge composite it is rescaled to 0-10 before its 15% weight applies.
The bonus for grabbing a player partly so your rivals cannot. If he is the last decent free agent in a thin position, taking him hurts everyone else's options — like buying the last sensible umbrella before a wet weekend.
Why it matters In a draft league players are one-of-a-kind, so denying a rival is real value.
Good number It carries 10% of the edge score; anything positive means the pickup also blocks somebody.
A news-aware dial on a player's projection. The news agent reads his latest bulletin and chance-of-playing and returns a multiplier: 1.0 is neutral, below 1.0 discounts a knock or a doubt, 0.0 is injured or out, and up to 1.2 gives a small bump for a positive story like a return to fitness. The projection is multiplied by it.
Why it matters It is how fresh news actually moves a number, instead of just sitting in a feed.
Good number 1.0 is neutral; 0.75 is a real doubt; 0.0 means do not expect him to play; 1.2 is the ceiling.
The chance the manager actually starts this player, judged by the rotation agent from his recent role, the club's rotation pattern over the last 8 gameweeks, and fixture context like a midweek European tie. It runs 0 to 1 and multiplies the projection, so a likely bench sees his number cut.
Why it matters A big projection is worthless if he is rested — this is the discount for that risk.
Good number 0.95+ is nailed on; around 0.6 is a genuine rotation risk; 0.0 is injured or out.
A points nudge for how well a player's style fits the next opponent, set by the tactical agent. Unlike NAF and MRP it is added to the projection, not multiplied: a favourable matchup adds points, a hostile one subtracts them, and most fixtures sit at 0.
Why it matters It captures the stylistic mismatch a raw average misses — a target man against a small back line.
Good number Runs -3.0 to +3.0 points; most fixtures are near 0, and any non-zero value comes with a stated reason.
Your place in the queue when two managers want the same free player — lower number goes first. This league runs Reset Weekly: every gameweek the order resets to reverse standings, so the bottom club always picks first and landing a player this week costs you nothing next week.
Why it matters Because priority resets, there is no reason to hoard it — claim when the player is worth it.
Good number Priority #1 wins every tie; league leaders live at the back of the queue by design.
The model's estimate of the chance a waiver claim actually goes through — that nobody with an earlier turn takes the player first and your drop is still there to make room. 95% is close to a formality; 40% means expect a fight.
Why it matters A brilliant swap you probably will not land is worth less than a decent one you will.
Good number 90%+ is near-certain; below 60% means plan for the fallback.
How many projected points a swap adds: the claim's projection minus the dropped player's, for the coming gameweek. This is the decision number — the engine ranks claims by it, gates the shortlist on it (at least +1.5 this gameweek, or +1.5 per gameweek averaged over the run-in), and the scorecard grades it against what actually happened. The edge composite is a comparison lens shown alongside; nothing is decided by it.
Why it matters When the brief says make a swap, this is the number it acted on.
Good number +1.5 is the bar to appear at all; +3 or more is a conviction call.
What a claim is worth once you account for the chance it fails: the points gain if it lands, multiplied by the odds it lands. A +6 swap with a 50% chance is worth +3 in expectation — the same as a guaranteed +3.
Why it matters It is the honest currency for comparing safe swaps against long shots — the queue optimiser's decision number, built on the same points gain the engine ranks by.
Good number The queue is ordered to maximise the total E[gain] across all your claims.
When a rival takes the player you queued before your claim is processed — either they queued him in an earlier slot than you, or in the same slot with a better waiver priority. Your claim then fails and your queue moves to its fallback.
Why it matters Queue order is a game of chicken; the optimiser exists largely to dodge snipes.
Good number A low snipe probability on your top claim is the goal — that is what a 90%+ land figure means.
Your waiver claims go in as a numbered list, and FPL processes them round by round: every manager's slot-1 claim is settled before anyone's slot-2 claim is looked at. So slot 1 is not just your favourite — it is the one that races every rival's first pick.
Why it matters Putting the wrong claim first can cost you a player a rival grabs one round earlier.
Good number Slots are processed 1, 2, 3... across the whole league — early slots are the valuable ones.
A backup claim on the same drop: if the main target is sniped, the fallback fires for the same outgoing player. If the main claim lands, the fallback is skipped automatically — you never end up dropping the same man twice.
Why it matters It is insurance: a failed claim becomes a decent pickup instead of nothing.
Good number Up to 2 fallbacks per primary; each must clear a raw gain of +2.5 to be worth a slot.
Three versions of the same waiver week, replayed after the fact. Naive is what was actually submitted. Optimised is what the model's reordering would have landed against what rivals really did. Optimal is the perfect-hindsight best ordering — a ceiling nobody could reliably hit in advance.
Why it matters Comparing the three shows whether the optimiser earns its keep, honestly.
Good number Optimised beating naive is the win condition; optimal only marks how much was on the table.
The points difference between the optimised queue and what was actually submitted, added up across the season. It can go negative in a given week: the optimiser bets on expected value, and sometimes the dice land the other way. Those weeks are reported as-is, never rounded up.
Why it matters It is the single number that says whether the queue reordering is worth trusting.
Good number Cumulatively positive is the target; single negative weeks are normal variance, not a fault.
How much rival demand there is for the players you want. The model predicts each opponent's likely claims from their habits — how often they act, what they chase after a loss — and uses that to price the risk on each of your claims.
Why it matters A free agent nobody else wants is safe at any slot; a contested one needs an early slot.
Good number High contention on a target pushes his land % down and his queue slot up.
Before recommending anything, the model puts your whole squad on trial — every player is a potential drop, not just the obvious weak link. For each drop slot it shortlists the top 2 candidate replacements, and a swap makes the brief if it is worth at least +1.5 points this gameweek or +1.5 per gameweek averaged over the run-in. A drop slot with any qualifying swap always keeps its best one, even when the overall cap is full.
Why it matters It catches the quietly declining player you have stopped looking at.
Good number 2 candidates per drop slot, top 12 overall (plus one guaranteed per otherwise-starved slot), +1.5 minimum gain to appear at all.
Every recommended swap is judged over the next few gameweeks, not just one — up to 4 ahead. A player with a kind run of fixtures can beat a bigger name whose next month is brutal, and the brief cites both the one-week and the run-in numbers. This is a decision number: a swap can make the shortlist on its run-in average alone, even when its next single gameweek is quiet.
Why it matters It stops you buying one good week attached to three bad ones, and stops a slow starter with a strong month being filtered out unseen.
Good number The window is capped at 4 gameweeks (or fewer near the season's end); +1.5 per gameweek averaged over it clears the shortlist bar.
A flag on a recommended claim whose value is mostly in the weeks ahead, not this one. When a player's run-in gain clearly beats his single-gameweek gain — a favourable run of fixtures keeps paying — the brief marks him 'ahead of the game': worth claiming now rather than waiting for the swing to arrive.
Why it matters It surfaces the pre-emptive buy — the player to get in before his good fixtures make him obvious to everyone else.
Good number Fires when the run-in gain is at least +3.0, exceeds this week's gain by +2.0 or more, and is at least 1.5x it.
A 0-to-1 score for how much the numbers and the written reasoning agree about a pick. 1.0 means the stats and the scouting story point the same way; low values mean they are arguing. Below 0.6 a HIGH-confidence pick is automatically demoted to MEDIUM — the model is not allowed to sound sure when its own evidence is split.
Why it matters It is the model's built-in humility check on every recommendation.
Good number 0.85+ is comfortable agreement; below 0.6 triggers the HIGH-to-MEDIUM cap.
Every recommendation carries one of three labels for how sure the model is, based on the strength of the underlying stats. HIGH is a conviction call, MEDIUM is a reasonable bet, LOW is a speculative punt. The label is a cap, not a promise — a HIGH pick gets knocked down to MEDIUM if the evidence disagrees with itself.
Why it matters The tier tells you how much of your own judgement to overlay on a pick.
Good number Check the calibration table: each tier's real-world hit rate is measured, not assumed.
Checking the model's confidence against what actually happened. Over a rolling 6-gameweek window, every scored pick is bucketed by its tier and the real points gains are averaged — if HIGH picks do not actually outperform, that is a problem worth knowing about.
Why it matters A weather forecast is only useful if 80% chance of rain means rain 8 times in 10.
Good number A tier needs at least 5 scored picks before its own numbers are trusted over the global average.
Of the swaps the model recommended, the share that actually gained points the following gameweek. A hit is any swap where the new player out-scored the dropped one — by 1 point or by 15, both count once.
Why it matters It is the simplest honest answer to 'is this thing actually right more often than not'.
Good number Above 50% means the model beats a coin flip; read it alongside average gain, not instead of it.
Re-running the recommender writes the same suggested swap to the database again, so the raw row count inflates. Before scoring, duplicates of the same gameweek-drop-claim combination are collapsed to one — the latest — so a swap suggested five times counts once, not five times.
Why it matters Without it the report card would grade the model on how often you pressed the button.
Good number Distinct swaps is the honest sample size; the raw count is shown only for transparency.
A yes/no gate on whether the calibration numbers mean anything yet. It flips to yes only when at least one tier's results are statistically distinguishable from zero — not just a handful of picks that happened to land. Until then the model treats its own track record as too thin to lean on.
Why it matters Small samples flatter and deceive; this stops the model grading itself on 4 data points.
Good number It is a real trust gate, not merely a tier reaching 5 picks — no signal means no self-adjustment.
Each gameweek you play one other manager: your starting XI's points against theirs, winner takes the fixture. You do not need the highest score in the league — just one more point than the manager opposite you this week.
Why it matters It changes strategy: beating your opponent's likely XI matters more than raw totals.
Good number The overview's projected scoreline is the week's whole story in two numbers.
A player no manager in the league owns. Unlike regular FPL, a draft league has no prices and no shared ownership — every player belongs to exactly one squad or to nobody, and free agents are the shopping aisle for waiver claims.
Why it matters The quality of the free-agent pool sets the bar every swap has to clear.
Good number A free agent projecting within a point of your worst starter is worth a claim.
The pre-season event where squads are built: managers pick players in turn, and the order reverses every round — whoever picks last in round one picks first in round two, like passing a plate both ways round a table. Build your board in the Draft Room before it starts.
Why it matters Your draft seat decides which pairs of players you can realistically land.
Good number Picking 1st means picks 1 and 16 of the first two rounds; picking 8th means 8 and 9 back-to-back.
A draft-day yardstick: how many points a player gives you over the best player you could pick up for free in that position later. A 6-point striker is not a bargain if 5-point strikers go undrafted; a 5-point goalkeeper is gold if the next best manages 2. The VORP column on the Draft Room board.
Why it matters It tells you which positions to draft early and which to leave until the late rounds.
Good number Draft the biggest gap over the free pool, not the biggest raw score.
FPL's own draft ranking, the order every rival sees in the official app. Autodraft picks straight down this list, so it is the best available read of what the room will actually do. Their opinion, not ours: it never moves our projections, and a player FPL has not ranked shows a dot, never a made-up number. When FPL's list does not cover the board's players, every rank is withheld rather than applied to a board it does not describe.
Why it matters Survive odds on the live board follow this list, because rivals draft from it, not from our model.
Good number A player our board loves but FPL ranks 100 or worse usually survives back to your next pick.
One draft list that mixes our model with FPL's, weighted by evidence. Each player's score is his VORP rank times his share of 900 minutes played last season, plus FPL's rank times the rest; the board then re-ranks everyone by that score. A full season means all our model, no minutes means all FPL's list. A player FPL has not ranked keeps our rank as his score, though FPL's opinion of OTHER players can still move him in the final order. Zero fitted parameters.
Why it matters It gives the no-evidence '?' tier an honest order instead of an arbitrary one.
Good number For anyone with 900+ minutes it simply equals our VORP order; differences only appear below that.
The chance a player is still undrafted when your next turn comes round. It counts the picks between now and your turn, assumes rivals mostly work down FPL's own list order, and allows for the fact that a manager who already has five midfielders is less likely to take a sixth. High is good: high means you can take someone else first and come back for him.
Why it matters It is the only number that answers the question that is actually hard on the clock: take him now, or take the scarcer player and get him next time round.
Good number 70% or more shows green and he is probably still there. 30% or less shows red and he is probably gone. Amber in between is a coin toss, and this estimate is not precise enough to break one. It is a rough estimate that has never been checked against real drafts.
A grade inside the player's own position, S at the top down to F. An S-tier defender is graded against other defenders, never against strikers. A player with under 270 minutes of Premier League evidence gets ? instead of a grade, and a player who has left the league gets X.
Why it matters The gap between the last B and the first C at a position is where waiting starts costing you. Tiers are how you see that gap before it opens.
Good number ? is not a bad grade, it is no grade: those players are ordered by FPL's list instead of ours. X cannot be drafted at all.
Season points, per position. The gap between the best player available at that position now and the best one likely to still be there at your next turn. It assumes rivals take the top of our value ranking, which they will not do exactly, so read it as a direction rather than a figure.
Why it matters It is the number that sets your draft order: take the position that costs most to wait on, leave the one that costs nothing.
Good number Under 1 point the room treats it as noise and says nothing. The position with the biggest drop is the one to take this round. May empty means the model expects nothing left there at all.
The points the model expects a player to score across the whole 26/27 season: last season's points per 90 minutes, pulled toward his position's average by how many minutes he actually played, multiplied by the minutes he is expected to get, with the official availability factor applied. An estimate built from last season's evidence, not a prediction.
Why it matters It is the raw material for VORP. On its own it is the wrong number to draft on: a defender's 140 and a striker's 140 are not the same asset.
Good number Read it against the other players on this board, not against a bar. This season total has never been checked against a played season, so there is no accuracy figure to quote and none is given.
The PICK line on the live draft page is needs-aware best value: the biggest value-over-replacement player at a position your squad still needs. We built two smarter 'pick edge' formulas and made each fight for the job in a simulated draft tournament (full 8-team snake drafts, every seat, 20 seeds, scored on the final squad's best legal XI). Both lost to the simple rule. One-step regret of waiting averaged 1435.4 points, a full-rollout planner 1464.8, needs-aware best value 1505.9; a 40-seed rerun agreed (1436.6 / 1463.9 / 1508.3), and each formula lost from all 8 draft seats. The clever formulas overweighted scarce positions against absolute value, so they are retired as diagnostics and the winner fronts the page.
Why it matters The pick you see on draft night is the strategy that actually won the simulation, not the cleverest formula we tried.
Good number needs-aware best value 1505.9 mean best-XI points vs 1464.8 (rollout) and 1435.4 (one-step), 20 seeds x 8 slots.
Last season's actual points minus what the same season was worth if every goal and assist had landed exactly on its expected (xG and xA) numbers. A big positive gap (HOT) means the player's shots went in more often than chances of that quality usually do; a big negative gap (COLD) means they went in less often. Both kinds of gap tend to shrink back toward normal the following season, so a HOT total is probably a little inflated and a COLD one a little unlucky. The chip appears at a gap of 20 points or more, either way.
Why it matters It flags which last-season totals to trust a little less, without changing any ranking: the number is display only.
Good number Within about 10 points of expected is normal; 20 or more either way earns the chip and some scepticism.
A player FPL still lists but who has left the league — sold abroad, loaned out for the season, or unattached (status 'u'). He stays visible on the board for honesty, struck through with a DEPARTED badge, but he is excluded from best available, scarcity depth, survive odds, the draft-order blend and every mock or tournament pool. FPL's own rank for him is shown but never drives anything: FPL ranked Son 12th while he plays in MLS.
Why it matters A board that quietly ranks players who cannot be drafted wastes picks; marking them keeps the pool honest.
Good number Zero departed players in any best-available list or depth count, ever.
A live measurement of which list the draft room is actually following. For every pick any manager makes, we check whether that player sat inside the top 5 of what was left under two orderings: FPL's own list and our value order. The share of picks that land inside each top 5 is the agreement score, and the page reads out which order the room tracks more. It says nothing until 8 picks are in, because a read on fewer is noise wearing a number.
Why it matters Survive odds assume rivals follow FPL's list; this line tells you whether tonight's room really does, measured rather than assumed.
Good number A score near 100% means the room drafts straight down that list; both scores low means the room is doing its own thing and survive odds deserve extra scepticism. Display only, it never moves a rank.
Some entries above describe parts of the full tool this hub does not serve, such as the weekly waiver brief and the queue optimiser. They stay here because the words turn up in the same places and a definition you can read is better than one you cannot. Every definition matches the model code; when the code changes, this page changes with it.
See the terms on the real board rather than in the abstract.
Open the boardNot affiliated with, endorsed by, or connected to the Premier League, Fantasy Premier League, or any club. Player data is derived from publicly accessible Fantasy Premier League endpoints and remains the property of the respective rights holders. Analytics, not betting.
This page last changed 2026-08-02.