What the archive is
The public archive is not a portfolio. It is the settlement log of analytical reports the terminal has produced for a curated set of leagues, kept online in full, with the final result stamped next to the original projection.
The two datasets, kept separate
- Internal historical dataset (14,000+ matches). Used to train, back-test and calibrate the model. Not published match-by-match — it is a research corpus, not a track record.
- Public settled reports (selected leagues). Since the platform went live, every analytical report produced for the selected leagues is published after settlement. The live counter on the archive page shows the current total; it grows every matchday.
Why we publish it
A football intelligence platform that hides its misses is not a research platform. Every settled report — matched or recalibrated — is what powers the next weight update in prediction_weights and the next calibration figure on the transparency board.
What's next
The next calibration release will introduce a per-league confidence band, computed from each league's own settled sample. The public archive is the source of truth for that release.