Data Sources
Data Sources and Model Boundaries
Key Takeaway
The site separates public fixtures and scores, prediction files, score distributions, public match-center lineup data, and AI summaries. Probabilities, xG, score calls, and risk labels are model-assisted analysis, not official outcomes.
Core Data Tables
Teams, dates, groups, host location, and completed scores once available.
Win-draw-loss probability, expected goals, knockout probability, and score-distribution inputs.
Group and knockout advancement probabilities, title probability, and finalist probability.
Starters, substitutes, formations, and player event stats appear only after the source publishes them.
Official FIFA squad-list PDF parsed into team player pools, key players, likely XI candidates, bench depth, caps, goals, clubs and position depth.
Public ESPN scoreboard odds fields are cached locally when a fixture is matched; provider names such as DraftKings are shown when supplied by the source.
The site attempts the public China Sports Lottery fixed-bonus feed, but keeps official odds marked pending when the source blocks automated access or has no matched fixture.
What Is Estimated
Win probability, expected goals, recommended scorelines, total-goals leaning, BTTS, upset risk, model fair odds, AI final forecast, and post-match model-review conclusions are generated from prediction files and page logic.
What Is Public Fact
Completed scores, fixture dates, groups, host location, published match-center lineup/player-event data, and cached public market odds are treated as public data. Official Sporttery odds are displayed only when the official source returns a matched fixture.
How AI Uses the Data
The AI forecast uses only provided model probabilities, score distributions, standings context, and public data summaries. It does not treat unavailable injuries, official odds, weather, referees, or private team news as facts.