GOSSIPBALL
Every term on this site

Guide to football stats and predictions

What each number means, how it's calculated and where the data comes from.

Match predictions

Home / Draw / Away chanceH · D · A, 1X2
GossipBall's chance of each result after 90 minutes plus stoppage time (extra time and penalties don't count). The three always add up to 100%. The coloured bar shows them side by side: green for the home team, grey for a draw, orange for the away team.Example: 55% · 25% · 20% means that in 100 matches like this one, the home team would be expected to win about 55.
Favourite
The result with the highest chance. A favourite with 45% still loses more often than it wins, so a favourite is the likeliest single result, not a certainty.
Clear favourite
A match where one result has a 60% chance or more. Last season these came in about 7 times out of 10, which is far better than average but still means roughly 3 in 10 went the other way.
Likeliest score
The single most probable final score. Even the likeliest score usually has only a 10–13% chance, because there are so many possible scores. It's often 1–1 even when one team is favourite, because a draw score can be the most common single scoreline while the favourite's wins are spread across 1–0, 2–0, 2–1 and so on.
Expected goals (prediction)
How many goals GossipBall expects each team to score in this match, based on both teams' attack and defence ratings and home advantage. It's an average: a team with 1.6 expected goals will score 0, 1, 2 or more on the day.This is GossipBall's forecast. It's different from match xG (below), which measures the chances a team actually created.
Scoreline chances grid
Every possible score from 0–0 to 5–5 with its chance. Rows are home goals, columns are away goals. Green squares are home wins, orange are away wins, grey are draws, and darker means likelier. The gold outline marks the likeliest score.
In play / Awaiting result
The match has kicked off. The prediction stays visible until the result arrives in the next daily update, when the match moves to Results.

Goals

3 or more goals (and other goal totals)
The chance that the match has at least a certain number of goals in total: 1 or more, 2 or more, 3 or more, and so on.Example: 58% for 3 or more goals means GossipBall gives a 58% chance of three or more goals, and 42% of two or fewer.
Both teams score
The chance that each team scores at least one goal.
Clean sheet
A team doesn't concede. The chance one team keeps a clean sheet is the same as the chance the other team scores zero.
Win without conceding
A team wins and doesn't concede, for example 1–0 or 3–0.
Team scores 2 or more
The chance a particular team scores two or more goals, whatever the other team does.
Winning margin
The chance of each goal difference: home by 1, 2 or 3, a draw, or away by 1, 2 or 3. Bigger wins (by 4+) are rare and not listed.

Match stats

Match stats
Numbers from finished matches, supplied by football-data.co.uk. Per game figures on league and team pages are season averages.
Expected goals (match xG)xG, HxG / AxG
A measure of the quality of the chances a team created. Each shot gets a probability of being scored based on things like distance and angle, and xG adds them up. A team with 2.1 xG created chances that would usually produce about 2 goals.xG is often a better guide to future performance than goals, which is why GossipBall blends it in.
xG againstxGA
The expected goals of the chances a team allowed its opponents. Lower is better.
Shots
All attempts at goal, including those blocked or off target.
Shots on targetOn tgt
Shots that would have gone in without a save by the goalkeeper (or a last defender on the line).
ConversionConv.
Goals divided by shots on target. A high number can mean clinical finishing or luck; it tends to drift back towards average over a season.
Corners
Corner kicks awarded to the team.
Fouls
Fouls committed by the team, as recorded by the officials.
Yellow / red cards
Cautions and sendings-off. A second yellow counts as a red in most data, and may also appear as a yellow.
Half-time goals
The score at half-time.
Referee stats
For leagues where referees are listed (England and Scotland), averages per game in matches they officiated this season: cards, fouls, goals and how often the home team won. Small samples make these noisy.

Tables and form

League table columnsP W D L GF GA GD Pts
P played, W won, D drawn, L lost, GF goals for, GA goals against, GD goal difference (GF − GA), Pts points (3 for a win, 1 for a draw). The table is calculated from results, so it won't include points deductions or some leagues' special rules.
xGF / xGA (table)
Total expected goals for and against this season. A team well above its xG may be riding good luck in front of goal.
Form
The last five results, oldest on the left: W win, D draw, L loss.
Clean sheets / Both score / 3+ goals (team stats)
The share of a team's matches this season with a clean sheet, with both teams scoring, and with three or more goals.
Head to head
The last meetings between the two teams in this league over the past five seasons.

Season odds

Season simulation
The rest of the season is played out 10,000 times. Each remaining match gets a random scoreline drawn from GossipBall's scoreline chances, points are added to the current table, and the final positions are counted. The share of simulations in which a team finishes first is its chance to win the league.Available for leagues where the whole remaining schedule is known. Leagues with play-offs or split seasons (Belgium, Turkey, Scotland) are not simulated.
Expected pointsExp. pts
The average final points total across all simulations.
Top 4 / Bottom 3
The chance of finishing in the top four or bottom three positions. What those positions mean (Champions League places, relegation, play-offs) differs by league.
Finishing position chances
For each team, the chance of finishing in each position. Darker squares are likelier; numbers are percentages (shown when 5% or more).

How GossipBall works

How GossipBall works
Each team gets an attack rating and a defence rating, estimated from five seasons of results, with recent matches counting more (a match from a year ago counts about half as much as last weekend's). For this season, GossipBall learns from a mix of 60% expected goals and 40% actual goals, because xG is steadier. Home advantage is estimated separately for each league. From the ratings it calculates each team's expected goals and turns them into the chance of every scoreline.
Poisson / Dixon–Coles
The maths behind the scoreline chances. Goals are treated as rare, independent events (a Poisson distribution) around each team's expected goals. The Dixon–Coles adjustment corrects the chances of low scores (0–0, 1–0, 0–1, 1–1), which happen slightly more or less often than the simple version predicts.
Attack / Defence / Overall rating
Ratings are on a log-goals scale relative to the league average (0). Attack +0.30 means scoring about 35% more than an average team against the same opponent. Defence +0.30 (shown as positive = better) means conceding about 26% fewer. Overall is attack plus defence.Newly promoted teams start near average and move as their results come in.
Home advantage
How many more goals home teams score than they would on neutral ground, estimated per league. Across our leagues it ranges from about 10% (Serie A) to about 30% (La Liga, Turkey, Scotland).
Time weighting
Older matches count less, with a half-life of about a year. This lets ratings follow changes in squads and managers without overreacting to one result.
European power ranking
Ratings from different leagues aren't directly comparable, so GossipBall adds a strength for each league, estimated from Champions League results and starting from a reasonable guess for each league. A club's European rating is its domestic rating plus twice its league's strength (once for attack, once for defence), with a small adjustment for its own European results.
League strength
How much better the average club in a league is than a club from a league we don't cover, on the same log-goals scale.
What the numbers can't see
Injuries, suspensions, team news, rotation, transfers, managerial changes, weather and motivation. It only knows results and expected goals, so treat its numbers as one informed view, not a certainty.

Track record

How the track record is testedwalk-forward test
To test fairly, GossipBall predicts each week's matches using only the results from before that week, then moves forward one week and repeats. Nothing from the future leaks into a prediction. Results pages show these pre-match predictions.
Favourite wonaccuracy
How often GossipBall's likeliest result was the actual result. About 50% is typical for football, because draws are hard to call and upsets are common.
Always picking the home team
A simple benchmark: predict a home win every time. It's right about 43–46% of the time.
Log loss
A score for the probabilities themselves, not just for who was expected to win. It punishes being confident and wrong. Lower is better. Guessing with only the league's usual home/draw/away rates scores about 1.08; GossipBall scores about 1.00.
Brier score
Another score for probabilities: the average squared gap between the predicted chances and what happened (1 for the result that occurred, 0 for the others). Lower is better.
Calibration
Whether the chances are honest: things GossipBall calls 30% should happen about 30% of the time. The calibration chart groups all predictions by their chance and shows how often they actually happened.

Data and updates

Where the data comes from
Results, match stats, expected goals and referees: football-data.co.uk. Season schedules: openfootball and football-data.org. Past Champions League results: openfootball. All are free sources. GossipBall does not collect, store or show bookmaker odds.
Update times
The site is rebuilt automatically twice a day. New results appear once the data sources publish them, usually within a day or two of the match.
Kick-off times
Shown in your device's time zone. TBC means the exact time hasn't been published.
Team names and badges
Clubs are named for identification only. The coloured squares with initials are generated by the site; no club or league logos are used.
Not a betting service
GossipBall is a football statistics site. It does not offer, promote or advertise gambling, shows no bookmaker odds and has no links to betting companies. Probabilities describe uncertainty; they are not tips or advice of any kind. See the legal notice.

GOSSIPBALL is a football statistics site, not a gambling service. We do not offer, promote, advertise or link to betting or gambling of any kind, show no bookmaker odds and receive no money from gambling companies. Predictions are statistical probabilities for information and entertainment only; they are not advice of any kind. Legal notice

Not affiliated with any league, club or governing body; names are used for identification only and no club or league logos are shown. Data: football-data.co.uk, football-data.org and openfootball. Photos: Sebastian Schuster, Victor, Lukáš Parničan and nika tchokhonelidze on Unsplash.

Updated 29 October 2026, 20:50 UTC · About · Legal & privacy · Guide · Track record · [email protected]