AIGOALDATA / Guide

How to read a prediction card

A card separates fixture identity, kickoff, market, choice, level and source. Pre-match estimates are never presented as actual final scores.

The header: which match are we talking about?

Every card starts with the identity of the match: the two teams, the competition, and the date and time of kickoff. The home team is always listed first. The time is shown as it was known at the moment of publication; a postponed match, or one whose kickoff time is uncertain, is not presented as a reliable prediction.

The card also shows when the analysis was recorded. This detail is essential: it proves that the reading was published before the match, and that it was not written with knowledge of the score.

Probabilities: estimates, not results

The probability block presents several scenarios: home win, draw, away win, both teams to score, and goal ranges such as “0 to 1 goal”, “2 to 3 goals” or “at least 4 goals”. The result probabilities (home, draw, away) complement each other and add up to 100%.

A probability is read over time. A scenario estimated at 70% should happen about seven times out of ten across a large number of comparable matches, and not happen about three times out of ten. Seeing a 70% scenario fail is therefore not an anomaly in itself; it is consistency over many matches that allows a judgement.

Goal ranges are read in the same way: they share the probability between closed matches, average matches and high-scoring matches. Their total also comes to 100%, which shows at a glance the expected profile of the game.

Expected goals and estimated scores

Expected goals give an estimate of how many goals each team could score, calculated before the match. They are often decimals, for example 1.4 for one team and 0.9 for the other: this is obviously not a possible score, but an expected average.

When estimated scores are shown, they indicate the scores considered most likely. Even the most likely score has, individually, a fairly low probability, because there are many possible scores. An estimated score must never be read as an announced score.

Decisions by market

For each market covered, the card shows a separate decision. “Selected” means a documented decision taken before the match. It can come with a level (A, B or C) and a direction, which refine the selection without ever turning it into a certainty.

“No bet” means that no selection is made on this market: the numbers did not justify taking a position. “Not evaluated” means that the data or the method did not allow a faithful evaluation. Neither status is a loss, and neither is counted as one in the results.

Each market is independent. A card can therefore show a selection on the straight winner, “no bet” on BTTS and “not evaluated” on another market. This should not be read as an inconsistency: each question gets its own answer.

Statistics and head-to-head meetings

The free card shows the teams’ key indicators and a summary of head-to-head meetings: number of games, last score, observed frequencies. VIP access adds detailed attack and defence statistics, split between home and away, the full head-to-head history and the list of matches used when it is available.

Head-to-head meetings provide context, but they are often few and sometimes old. They are presented separately, as a descriptive reading, and must not be confused with the model’s probabilities.

After the match: the result and its verification

Once the match is over, the final regulation-time score is added to the card. It first carries the status “reported”, then “verified” when a second reading or a review of evidence has confirmed it. Only verified results count towards the Performance page.

The pre-match analysis, for its part, does not change. Probabilities, decisions and levels remain exactly as they were published, so anyone can compare what was announced with what happened.

A complete reading example

Take a fictional card. The match brings together two mid-table teams on Saturday at 3 pm. The card shows a 48% probability of a home win, 27% for a draw and 25% for an away win, as well as 58% for both teams to score. Expected goals are 1.5 for the home team and 1.1 for the visitor.

On the decision side, the card shows a selection on BTTS, “no bet” on the straight winner, and “not evaluated” on another market for lack of sufficient data. Reading: the model sees a match in which both teams have a good chance of scoring, but no winner clear enough to justify taking a position on the result. After the game, a verified 2-2 would land the BTTS selection, without changing any of the other statuses.

The most common reading mistakes

The first mistake is reading the highest probability as a tip: a home win at 48% remains less likely than “draw or away win” combined. The second is taking an estimated score for an announced score. The third is thinking that no bet means something was overlooked or not analysed, when it actually reflects a decision.

Finally, a method should not be judged on a single weekend. A few losing, or winning, selections say almost nothing about the quality of an approach; only the Performance page, built on verified results and larger samples, allows a serious evaluation.

What a card is not

A prediction card is a statistical analysis, not a personal betting recommendation. It does not take into account your situation, your budget or your tolerance for risk. Betting can lose you money and is forbidden to minors. Set time and money limits, never try to “win it back” after a loss, and see our Responsible gambling page if needed.

Predictions Home