Types of data used
The core assessment includes verified completed matches, goals scored and conceded, venue, competition context and recent performance. Head-to-head records provide secondary context when available.
Relevant recent fixtures are considered across competitions rather than relying on one tournament alone, while differences in competitive level remain part of the context.
Missing or conflicting information
A missing source record does not prove that a team has played no matches. Identity matching, coverage and connection issues are considered separately.
When enough recent matches cannot be verified, confidence is reduced or the prediction is withheld. Missing values are not filled with zero as if they represented real performance.
- Incomplete matches are excluded from form
- Teams are matched with persistent identifiers
- Old H2H data receives limited weight
- Insufficient coverage is disclosed
Model outputs
Match result, total goals, both teams to score and score scenarios are produced within one probability framework. The main scenario is a score frame supporting the broader match view; the most likely exact score is one individual combination in the distribution.
Model confidence is not a result guarantee. It summarises data consistency and the clarity of the signal.
Limits and responsible use
Injuries, late line-up changes, red cards and in-game decisions can alter a match. A model cannot know future events that were not verified at publication time.
GoalReader analysis is not investment advice and does not replace personal judgement. Content should be used for information and entertainment.