Every analysis on aiALYSIS Live is made of the same four parts. Knowing them is the quickest way to understand what an analysis says — and what it does not. The examples below are illustrations; you can see real analyses every day on the Today's AI match analysis page.
The four parts of an analysis
Example: Total goals: 2.5 OVER — confidence 68%.
- Category says what is measured: total goals, first-half goals, home or away team goals, corners, cards, match result.
- Line (2.5) is the threshold used for the comparison.
- Direction (OVER/UNDER) says whether the actual value will end up above or below the line.
- Confidence (68%) shows how likely the model considers this scenario.
What do the line and direction mean?
“Total goals: 2.5 OVER” is the scenario of 3 or more goals in the match; “2.5 UNDER” means 2 or fewer. Because the line is a half number there is no in-between outcome: the analysis is resolved as either a hit or a miss. “Home team goals” counts only the home side’s goals and “away team goals” only the visitors’; “1st-half goals” covers only the first half and “2nd-half goals” only the second. “Match result” analyses have no line and are written with the team name (e.g. “Liverpool wins”).
How to read the confidence percentage
Confidence is a probability estimate, not a certainty. An analysis at 70% will sometimes miss; that is not a malfunction but the nature of probability. Confidence is tested across many analyses, not one match: how often analyses actually hit is shown in the public count on the aialysis.live/accuracy page. Displayed confidence is capped at 95%, and only analyses above a confidence threshold are published, so the published set is a selection, not every possible scenario.
Pre-match vs live analysis
Pre-match analyses are produced before kick-off from form, head-to-head history and season statistics. Live analyses also use in-play statistics such as minute, score, shots, corners, cards and tempo, and update when the score or situation changes. When a line is passed live (e.g. a 2.5 OVER analysis with the third goal), the analysis is resolved immediately. The two types are counted separately on the accuracy page.
What is the AI summary for?
The short text on the match page and on the daily analysis page makes a match’s analyses readable. It is based on statistics; no number that the model did not produce can enter the text. The percentages in the summary are the same as the confidence values in that match’s analysis rows.
Five things to keep in mind
- One match is not evidence. A single hit or miss does not show on its own whether a model is good or bad.
- Look at the category. Accuracy differs by category; the breakdown is on the aialysis.live/accuracy page.
- Read the sample size. A percentage based on few analyses can swing widely; the 30- and 90-day windows are more meaningful.
- Remember when it was published. A pre-match analysis reflects the data at the time it was published; later news (e.g. a late line-up change) may not be reflected.
- It is information. Analyses are statistical estimates, not certainties, and are not advice for any financial decision.