Glossary

The number of FPL points a player is expected to score in the next gameweek.

The feed is frozen at each deadline and graded against the real scores afterwards. Every table that shows it names its source: either FPL's own expected points or this site's model, whichever currently holds the feed. It is a ranking aid, not a promise.

See also: ep_next, Shadow model, Projection horizonxPoints tablesHow it is made

FPL's own expected points for a player in the next gameweek, published in its public data.

It leans heavily on recent form, so early in a season it rests on very few games. The site uses it as the baseline every model has to beat, and as the live feed until one does.

See also: xPoints, Promotion gate

A projection model that runs alongside the live feed without being shown as the feed, so it can be graded before anyone relies on it.

Its numbers are frozen at the same deadline as the feed's and scored against the same results, which makes the comparison fair.

See also: Promotion gate, Rank matchAccuracy

The rule that decides when the shadow model replaces the live feed: it must beat the feed over the last six gameweeks in which both were frozen, by more than chance.

The gate is checked after every gameweek and its verdict, with the reasons when it fails, is printed on the Accuracy page.

See also: Shadow model, Rank match, Top 20Accuracy

Expected points for each of the next three to eight gameweeks, not only the next one.

The next gameweek's number is split into a neutral scoring rate and the things that change week to week (expected minutes, the opponent, home or away, a return date, a blank or a double), then put back together for every gameweek ahead. Later weeks lean less on the feed and more on what a player of that price usually scores.

See also: xPoints, xMins, Opponent factorTransfer PlannerHow it is made

How much a fixture raises or lowers a projection, from the opponent's attack or defence by expected goals this season.

Attackers are scaled by the defence they face and defenders by the attack. Home games count a little more and away games a little less. Only half of the opponent's strength is credited, because the feed already reflects the next fixture in part.

See also: FDR, Projection horizon

A haul is 10 points or more in a gameweek, a return is 6 or more, and a blank is 2 or fewer.

The Captain Picker simulates each candidate's gameweek several thousand times and reports the chance of each, with a floor (the 10th percentile) and a ceiling (the 90th).

See also: Captaincy simulationCaptain Picker

A simulation of each captain candidate's next gameweek from his own rates: goals and assists from xG and xA per 90, bonus, cards, defensive contributions and his team's expected goals conceded.

This season's rates are blended with last season's by minutes played, thin samples are pulled towards a typical player in the position, and the result is scaled by minutes, fixture difficulty and availability, then scored with FPL's published rules.

See also: Haul, return and blank, Effective ownershipCaptain PickerHow it is made

The minutes a player is expected to play in the next gameweek.

It comes from his last six gameweeks, the latest counting most, with a mild prior early in the season, multiplied by FPL's own availability. A double gameweek adds both games and a blank is zero. An amber number means the chance of sixty minutes is under half.

See also: P(start), P(60+), Role, Chance of playingHow it is made

The chance a player starts at least one of his club's games in the gameweek.

See also: xMins, P(60+)

The chance a player plays sixty minutes or more, the line at which FPL pays the second appearance point and defenders and keepers qualify for a clean sheet.

The served figure averages the site's own estimate with the xPoints model's, and player pages show both halves. It is graded after every gameweek with a Brier score, the blend and each half.

See also: xMins, Brier score

A word for how secure a player's place is: nailed, likely starter, rotation, bench, fringe or out.

Nailed means he starts at least 85% of games when fit, likely starter 60%, rotation 35%. Bench means he rarely starts but often comes on. Out means FPL gives him no chance of playing, or his loan terms bar him from the fixture.

See also: xMins, Loan ineligible

FPL's published estimate that a flagged player is fit for the next round: 75%, 50%, 25% or 0%.

The site multiplies the next gameweek's projection by it. For later gameweeks it reads the return date in FPL's news instead; a player listed as out with no date is treated as out for the whole horizon.

See also: xMins, Projection horizonTeam news

A loaned player who cannot face his parent club. FPL flags the gameweek, and the site projects zero for him that week.

It shows on the player's page under his status and in his fixtures, and the planner, optimiser, chip advisor and captain picker all treat that gameweek as a blank for him.

See also: Blank gameweek, Role

FPL's form: a player's average points per game over the last 30 days, whoever the opponents were.

See also: Fixture-adjusted form, Form momentum

Form over the last five gameweeks, re-weighted by the strength of the opponents the points came against.

Each game's points above the appearance points are scaled by the opponent: the defence faced for midfielders and forwards, the attack faced for keepers and defenders. The scaling is capped so one lopsided week cannot dominate. It separates players who score against everyone from players who fed on weak opponents.

See also: Form, Opponent factorHow it is made

Whether a player's recent scoring is speeding up or slowing down: accelerating, rising, stable, declining or collapsing.

It compares his most recent games with the ones before. From the analytics warehouse, available from Gameweek 1.

See also: FormPlayer tables

Points a gameweek a player gives over the freely available alternative in his position.

Replacement level is the median of the best eight playing options at or below the floor price: 4.5m for keepers and defenders, 5.5m for midfielders, 6.0m for forwards. A second figure compares him with players within half a million of his price.

See also: PAR per millionValue tableHow it is made

PAR divided by the price above the floor for the position: whether a premium is buying points.

It is not given for players within a few tenths of the floor, where a tiny denominator would make the figure meaningless.

See also: PAR

How steady a player's gameweek scores are: the spread of his points relative to his average.

A low figure with a high average is a reliable scorer; a high figure means his points come in bursts. From the analytics warehouse.

Consistency table

A player few managers own. On the Differentials page, under 5% owned is an ultra differential and 5% to 15% a differential; gameweek pages use under 10%.

The differential score ranks low-owned players on form, expected goal involvements, the fixtures ahead and set-piece duty.

See also: Effective ownership, xGIDifferentials

Expected goals: the goals a player's shots would produce on average, given where and how they were taken. FPL publishes it per match.

See also: xA, xGI, xGI diff

Expected assists: the assists a player's passes would produce on average, given the chances they created.

See also: xG, xGI

Expected goal involvements: xG plus xA, the goals and assists a player's chances were worth.

See also: xG, xA, xGI diff

Actual goals plus assists, minus xG plus xA. Below zero, a player is scoring less than his chances suggest; above zero, more.

A large gap in either direction tends to close over time, so a big negative number is often a player about to score.

See also: xGIxG table

Expected goals conceded while a player was on the pitch. For defenders and keepers, lower is better.

See also: xG

FPL's Bonus Points System scores every action in a match. The top three BPS scores in each match earn 3, 2 and 1 bonus points.

The Live Tracker and the gameweek pages work out provisional bonus from live BPS, with FPL's tie rules, until FPL confirms it.

Live Tracker

Two FPL points for defensive work in a match: 10 clearances, blocks, interceptions and tackles for a defender, or 12 including recoveries for a midfielder or forward.

Player pages show defensive contributions per 90 and how often the threshold is reached.

FPL's own index combining Influence, Creativity and Threat into one number per player.

Where a player ranks among players in his position with enough minutes, per 90 minutes this season. 100 is the best.

A player below the minutes threshold still gets his own numbers, with a note that the sample is small.

FPL's Fixture Difficulty Rating for each game, from 1 (easiest) to 5 (hardest), from the side of the club in question.

Beside it the site shows where the opponent's attack or defence ranks among the 20 clubs by expected goals this season, which moves faster than FPL's rating.

See also: Opponent factorFixture Planner

A gameweek in which a club plays twice. Projections and expected minutes add both games together.

See also: Blank gameweek, Chips

A gameweek in which a club has no fixture. Its players project zero that week.

See also: Double gameweek, Chips

A finished gameweek is provisional until FPL has checked the data and added bonus, then final. Points can still change while it is provisional.

Gameweeks

The share of all FPL managers who own a player, as FPL publishes it.

It includes abandoned teams, so it understates how many active managers hold a popular player.

See also: Top 10k and top 100k ownership, Effective ownership

Ownership among the best-ranked managers, estimated from a sample of their teams each gameweek, with the sample size shown.

It answers what engaged managers hold, which overall ownership cannot.

See also: Ownership, Template

The players most serious managers own at a given time. Holding the template protects rank; moving away from it is how rank is gained.

See also: Top 10k and top 100k ownership, Differential

Ownership plus captaincy: each owner counts once, each captain twice and each triple captain three times, as a share of managers.

Above about 75%, a haul moves you with the field even if you own him; below about 35%, it moves you against it. FPL does not publish captaincy before the deadline, so the Captain Picker estimates it and says so. Inside a mini-league it is exact, from the managers' picks.

See also: Ownership, Captaincy simulationCaptain PickerMini-League

One transfer a gameweek costs nothing. An unused one rolls over, and up to five can be banked.

The site works out how many you hold from your public history. The limit is read from FPL's published rules.

See also: HitTransfer Planner

Four points deducted for each transfer beyond your free ones.

The optimiser and planner charge the hit against the gain and show how many gameweeks the gain would have to hold for it to pay back.

See also: Free transfer

What you get for a player: his purchase price plus half of any rise since, rounded down to 0.1m. A fall is passed on in full.

The site reads purchase prices from your public transfer list. Players from your opening squad are taken as bought at their season-start price.

See also: Price change

FPL moves a player's price by 0.1m overnight when net transfers push its own tracker to plus or minus 100%.

FPL publishes each player's progress, his current rate and its own projection for the next three nights. Price Predictions reads those rather than guessing from transfer counts, and shows players who are locked (new signings cannot change for about two weeks).

See also: Selling pricePrice Predictions

One-off boosts. Wildcard: unlimited free transfers that stay. Free Hit: unlimited transfers for one gameweek, then the squad reverts. Bench Boost: all fifteen players score. Triple Captain: the captain scores three times.

Each chip can be played once in each half of the season. The Chip Advisor scores every remaining gameweek for each chip against a typical week, so the question it answers is when, not whether.

See also: Double gameweek, Blank gameweekChip Advisor

The average distance between projected and actual points, in points. Lower is better.

Most players score nothing in a given week, so the Accuracy page sets it beside the score for predicting zero for everyone. A feed that cannot beat that is not telling you much.

See also: Rank matchAccuracy

Whether the feed put players in the right order, among those who started. 1 is a perfect order, 0 is no better than chance.

Order matters more than the exact number when you are choosing between players. It is shown with a 95% range, because one gameweek is a small sample.

See also: MAE, Top 20Accuracy

The share of the feed's top 20 players who finished in the real top 20 that gameweek.

See also: Rank match

The best score any player got in a gameweek minus the score of the feed's top pick. Lower is better, and zero means the feed named the week's best player.

See also: Top 20, Captaincy simulationAccuracy

A score for probabilities: the average squared gap between the chance given and what happened. 0 is perfect, and always saying 50% scores 0.25.

The site uses it to grade the chance of sixty minutes.

See also: P(60+)

Projections are saved at each deadline and never edited afterwards, so the grade is against what was said before the games, not a later revision.

How grading works