The current radar estimate is a transparent heuristic intelligence score. It is not a guaranteed forecast, not personalized advice, and not a trading instruction.
RADAR ESTIMATE starts from the market probability and adjusts for 24h probability movement, volume, liquidity, matched news, and crypto impulse when relevant. The result is a research estimate used to detect divergence from the market, not a promise that the radar is right.
If a read has no external source match, it is labeled market-data-only. If the worker is fresh but a source is failing, the desk shows DEGRADED. If follow-up data is not old enough, the ledger shows Pending instead of claiming accuracy.
These components explain why a read appears higher or lower on the desk.
Distance between market-implied probability and the heuristic estimate.
Recent probability movement from the market price feed.
Depth and flow used to reduce thin-market false positives.
Matched external news increases evidence score; market-only reads stay under calibration.
Only real CLOB probability history earns a freshness contribution; unavailable history is not synthesized.
Thin liquidity and weak evidence reduce the final signal score.
Trust comes from review, not loud claims.
Each tracked read records the opening market probability and radar estimate when the worker sees it.
Reads are reviewed after 1h, 6h, or 24h depending on the event horizon.
Accuracy-style summaries should wait until enough live samples have accumulated.