AMD confirmation strengthened
Signal engine
Cross-source confirmation moved ahead of general news pickup, which raises confidence without removing risk.
Alert Center
Alerts are framed as research prompts, not trade commands.
The alerting model is built around confidence changes, contradiction buildup, and coverage gaps so users know why attention is required.
Alerts3
Open items (demo)
Current alert feed
Connect a database and add watchlist items with thresholds to see real alerts.
elevated
AAPL contradiction stack is widening
Cross-source confirmation
Narrative drift is visible, but broader search and filing support remain soft enough to keep confidence capped.
Gaming narratives need cleaner release timing
Tradability model
Community conviction is improving, but tradability remains constrained when release windows are still fuzzy.
Protocol
Alert design principles
The prototype makes confidence changes obvious without pretending the engine can predict outcomes.
Alerts fire once per threshold crossing. They reset when momentum drops below the threshold, so they can fire again on the next spike.
Watchlist alerts are shown as badges on the dashboard. No push notifications for a personal tool.
Pipeline errors are logged but don't block other sources. A Reddit failure still lets EDGAR and Trends run.