How ExtraTips football predictions work

What ExtraTips predictions mean

ExtraTips predictions estimate the chance of different outcomes in a football match. The model uses the match data available at the time. If Home is shown at 47%, the model estimates a 47% chance of a home win.

A red card, a deflection, a missed penalty or an exceptional finish can change the result of a match. To assess the estimates, compare the predicted percentages with how often those outcomes occur across many matches.

In the 47% example: the draw and away win account for the remaining 53%.

How a prediction is created

The process starts with structured football data. Depending on the competition and the information available, that can include recent results, goals scored and conceded, home and away performance, team strength, head-to-head history and other competition-level indicators.

A statistical model looks for relationships in the data and estimates the chance of each outcome. The written analysis explains those estimates and the patterns behind them.

Predictions for upcoming fixtures can be updated as the underlying data changes, so percentages may vary from one day to another. Missing or outdated data can affect the estimates.

How to think about 50%, 70% and other percentages

A 70% prediction assigns a 70% chance to an outcome. Across a large number of such predictions, that outcome should occur about 70% of the time if the estimates are accurate. This agreement between predicted chances and observed results is called calibration. The outcome would fail to occur about 30% of the time, with variation from one sample to another.

A single result provides too little evidence to assess calibration, and a sample of ten matches can vary considerably by chance. Track larger groups of predictions with similar percentages and compare them with the results.

How to read every ExtraTips prediction type

The match panel shows probabilities for several markets. When a set includes every possible outcome and only one can occur, its probabilities should total 100%. Displayed values may differ slightly because of rounding.

Full-time result (1X2)

This is the result after normal time, including added time but excluding extra time and penalties: 1 means a home win, X a draw and 2 an away win. In a 50%–25%–25% example, the home win is the single most likely result, but the combined chance of the home team not winning is still 50%.

Over/Under total goals

These lines use the total goals scored by both teams. A total cannot equal a half-goal threshold, so it falls either above or below the line. Over 1.5 means at least 2 goals; Over 2.5 means at least 3; Over 3.5 means at least 4. Under 2.5 covers 0, 1 or 2 total goals.

Both teams to score (BTTS)

Yes requires each team to score at least once. A 1–1, 2–1 or 3–2 result qualifies; 0–0, 1–0 and 0–2 do not. It does not ask which team wins or how many total goals there are beyond that condition.

First-half winner

This is another three-way market, but it stops at half-time. Home, Draw and Away refer only to the score when the first half ends. A team can lead at half-time and still draw or lose the match.

Half-time / full-time (HT/FT)

This combines the result at half-time with the result at full-time, giving nine combinations. Home/Home means the home side leads at half-time and wins; Draw/Away means a draw at half-time and an away win at the end. Read the row first (half-time), then the column (full-time).

Correct score

The grid assigns a probability to each exact final score. Home goals run down the rows and away goals across the columns, so the row labelled 2 and column labelled 1 means 2–1. A 2–1 result is one of several scores covered by a home win, and one of several covered by Over 2.5. In a consistent model, its probability cannot exceed the probability of either of those broader outcomes.

Written analysis and “Our picks”

The text explains the model estimates and may add context about recent form, playing styles or previous meetings. Read it alongside the percentages to see how much probability the model assigns to each outcome.

What fair odds mean

ExtraTips may show fair decimal odds beside a probability. They are calculated as fair odds = 1 ÷ probability, using the probability as a decimal. For example, 50% is 0.50, so 1 ÷ 0.50 = 2.00. A 40% estimate corresponds to 2.50, and 25% to 4.00.

Fair odds exclude the bookmaker’s margin. Bookmaker prices may also reflect market movement, risk management and information absent from the model. Any comparison with those prices depends on the accuracy of the model estimate.

What a model may not fully capture

A model may have incomplete or outdated information about injuries, squad rotation, tactics, travel or weather. Red cards and other events during the match can also change the result. Head-to-head records may be less relevant after changes to managers or squads.

Check the fixture time, market label and latest team news when reading the percentages. If you bet, keep to your budget and never increase a stake to recover an earlier loss.

How to evaluate predictions

Assess both the proportion of successful picks and the probabilities assigned to them. Hit rate counts successful picks. Calibration checks whether outcomes assigned 70% occur about 70% of the time across many matches, and those assigned 50% about half the time. These measures answer different questions about the predictions.

Use a large sample and compare predictions for the same market and time period. Record the original pre-match probabilities and include every prediction that meets the criteria set before the review, including unsuccessful ones. Short winning or losing streaks provide limited evidence.

Responsible use

ExtraTips predictions are provided for information and entertainment. They are not financial advice, betting advice or a promise of profit. If you choose to bet, set a budget before you start, never chase losses and use the control tools available to you. Visit Responsible Gambling for help and practical guidance.