Glossary

A plain-English guide to the expected-goals, pressing and model metrics used across xG Lab. New to xG? Start at the top; jump to a term using the links below.

Expected goals (xG)

Expected goalsxG: The probability that a shot becomes a goal, based on factors like distance, angle and situation. Summing a team's shots gives how many goals an average finisher would have scored from those chances.On xG Lab: Used across match pages, team and player tables, and as the basis for our strength model.
Non-penalty expected goalsnpxG: xG with penalties removed. Penalties are worth a fixed ~0.76 xG and say little about open-play creation, so npxG is a cleaner measure of a team or player's chance quality.
FinishingnpG−npxG: Non-penalty goals minus non-penalty expected goals. Positive means a player has scored more than the quality of their chances suggests (clinical, or a hot streak); negative means they have been wasteful or unlucky in front of goal.On xG Lab: Shown as the Fin column on the Players table, computed from public goals and npxG.
Expected goals on targetxGOT: A post-shot model: given where a shot ended up on (or off) goal, how likely it was to be scored. The gap between xGOT and xG captures shot placement and finishing.
Expected assistsxA: The xG of the shot that a pass set up. It credits the creator for the quality of chance they created, whether or not it was scored.
Key passes: Passes that directly lead to a shot. A volume measure of chance creation that pairs with xA's quality measure.

Build-up & pressing

Expected goals chainxGChain: The total xG of every possession a player was involved in — any possession where they touched the ball that ended in a shot. It rewards players who are part of dangerous moves, not just the final pass or shot.On xG Lab: Highlighted for deep-lying playmakers and full-backs who don't rack up goals or assists.
Expected goals buildupxGBuildup: xGChain with the shooter and the assister removed. It isolates pure build-up contribution — the players who move the ball into dangerous areas before the final two actions.
Deep completionsdeep: Passes completed close to the opponent's goal (roughly within 20 metres, excluding crosses). A proxy for how often a team works the ball into threatening areas.
Passes allowed per defensive actionPPDA: How many passes the opponent is allowed before a defensive action (tackle, interception, foul) in the attacking three-quarters of the pitch. Lower PPDA means more aggressive, higher pressing.On xG Lab: Plotted on the Teams page pressing & control chart.

Outcomes & value

Expected pointsxPts: The points a team would expect from a match given the xG of both sides, simulated across possible scorelines, then summed over a season. It estimates where the table 'should' be on chance quality.
Expected goal differencexGD: Expected goals for minus expected goals against. A single number for how much a team out-creates its opponents.
Points over expectedPts − xPts: Actual points minus expected points. Positive means a team is outperforming its underlying numbers (good finishing, goalkeeping, or luck); negative means the opposite.On xG Lab: Shown as the Δ column and over-performance chart on the Teams page.

Model & predictions

Win probability: The chance of a home win, draw or away win at any moment of a match, updated minute by minute as goals, red cards and time remaining change the picture.On xG Lab: The win-probability replay on every match page.
Scoring rateλ: Each team's pre-match expected goals for the game, derived from their attack and defence strength. λ feeds the Poisson model that produces win probabilities and predicted scorelines.
Rolling strength: Attack and defence ratings built from a team's recent xG using an exponentially-weighted moving average that carries across seasons. Recent form counts for more than old form.On xG Lab: Drives the Power rankings and all pre-match predictions.
Out-of-sample: A prediction made without the model ever seeing that match's result. All pre-match forecasts on the site are out-of-sample, so the accuracy numbers reflect genuine forecasting, not hindsight.
Log-loss & Brier score: Two scores for the quality of probabilistic forecasts. Both reward confident, correct predictions and punish confident, wrong ones. Lower is better for each.On xG Lab: Reported on the Predictions and model pages.
Calibration: Whether predicted probabilities match real-world frequency: of all the games called a 70% home win, about 70% should actually be home wins. A well-calibrated model can be trusted at face value.
Monte Carlo projection: Simulating the remaining fixtures thousands of times from the strength model to estimate each team's chance of winning the title, finishing top-four or being relegated.On xG Lab: Powers the Projections page (title / European / relegation odds).
Model vs market: A comparison of the model's log-loss against the market's consensus closing forecast — the sharpest, most efficient benchmark available. We don't claim to beat the consensus; we show how close the model gets with its logic fully open.On xG Lab: The Model vs Market card and cumulative chart on the Predictions page.
Over/Under 2.5: The probability that a match's total goals is 3 or more (Over) versus 2 or fewer (Under), derived from each team's λ with a Poisson model.
Both teams to scoreBTTS: The probability that both teams score at least once, derived from the two teams' scoring rates (λ).

Reading the stats

Per 90/90: A stat scaled to a full 90 minutes played, so a regular starter and a substitute can be compared fairly regardless of minutes.
Percentile: A player's rank against positional peers on a 0–100 scale. A 90th-percentile xA means only 10% of comparable players create more.On xG Lab: The percentile bars and radar charts on player pages.
Squad changes: Arrivals and departures inferred by comparing who appeared for each club this season versus last. It is not a transfer database: there are no fees, dates, or loan-vs-permanent distinction, and moves to or from another league show one side only. Promoted clubs show only moves between clubs already in this league.On xG Lab: The Squad page (Ins & Outs and production turnover).
Production turnover: How much on-pitch production a club lost and replaced over the off-season, measured in xG + xA from the prior season. Departures count the production leaving; arrivals count what measurable signings produced before joining (players coming from outside the league can't be measured and are flagged separately).On xG Lab: The turnover line on each club's card on the Squad page.