Research Metrics

Combined research view for current production issues, EntryScore validity, and first-candidate quality. These metrics are explanatory, not runtime-authoritative.

home · dashboard · research-metrics · console · slugs · trades · diagnostics · twitter · github

Open Issues

Second retry live window capture is still weaker than target and can miss short recovery windows.
Partial pair completion and residual mismatch semantics still need cleaner production classification.
Some read paths still depend on tail parsing instead of indexed historical storage.

Research Metrics

EntryScore PG validity pipeline is live. These numbers are for research and audit, not runtime authority.

Pipeline

Live

Candidates

Good Outcome Rate

20.59%

Bad Outcome Rate

58.82%

Avg PnL Ratio

0.0061

Turnover-Adjusted Return

0.000282

Distinct Slugs

Completed Rounds / Slug

2.62

Automation is now complete: candidate validity no longer needs to be inspected slug by slug by hand.
Current quality is still weak: bad outcomes remain much higher than good outcomes.
Next focus is hourly and slug-level decomposition to find which windows still pull the strategy away from clean arbitrage behavior.