Join — $29

The Accuracy Lab

Every projection should leave a receipt. SLAITIQ tracks what it projected, when it changed and what actually happened. The goal is not to advertise perfection. It is to make the record inspectable.

Accuracy Lab is building its public record now. Live projection counts will appear here when the application database produces them automatically. No placeholder totals.

Public methodology

  • Timestamp everything. Preserve the first published projection and each material update.
  • Never silently rewrite history. Current projections can update, but previous values remain in the audit trail.
  • Attach actual results after the slate.
  • Publish misses. Do not surface only high-performing calls.
  • Separate pre-lock from post-lock. No projection is retroactively improved after lock.
  • Label model versions when methodology changes.
  • Show sample size with every accuracy statistic.
  • Avoid fake precision.

TOTAL PLAYER PROJECTIONS TRACKED

Will populate from tracked slates. Not published until sample size exists.

TOTAL SLATES TRACKED

Will populate from tracked slates. Not published until sample size exists.

TOTAL PROJECTION UPDATES LOGGED

Will populate from tracked slates. Not published until sample size exists.

MEAN ABSOLUTE ERROR (MAE)

Will populate from tracked slates. Not published until sample size exists.

MEDIAN ABSOLUTE ERROR

Will populate from tracked slates. Not published until sample size exists.

POSITION-LEVEL MAE

Will populate from tracked slates. Not published until sample size exists.

Open the live desk