Structured football data
Fixtures, team identities, schedule changes and market reference data are normalized into deterministic research inputs.
Duringa Craft is developing Football Intelligence Engine — an AI-assisted football research and decision-support platform built around structured data, historical signals, reference market information and reproducible evidence.
The system is designed to make football analysis inspectable rather than opaque. Data quality failures are recorded explicitly, source evidence is preserved, and research outputs are intended to be reproducible and reviewable.
Fixtures, team identities, schedule changes and market reference data are normalized into deterministic research inputs.
Source snapshots and data-quality states are preserved so important analytical decisions can be traced back to their inputs.
Large language models are being evaluated for research synthesis, anomaly investigation, explanation and natural-language interaction.
The project is being built with a controlled validation process: deterministic team mapping, explicit unmatched-fixture states, frozen evaluation universes, versioned schedule changes and isolated test accounting.
Unmatched or invalid observations are surfaced as explicit states instead of disappearing from the research sample.
Evaluation criteria are defined before a trial begins so results can be interpreted against pre-registered thresholds.
Provider-specific team identities are reviewed and approved to avoid runtime fuzzy matching in critical research workflows.
Football Intelligence Engine is not presented as a finished consumer product. The current focus is validating data acquisition, identity resolution, evidence retention and research methodology before broader productization.
Founder contact for technical, partnership and startup-program inquiries.