Get results of a scan batch created by signal_scan. Poll this until all scans complete.

Returns completed/pending/failed counts and results grouped by long_signal / short_signal / no_signal / condition_never_fired (the last lists symbols whose condition is valid but never evaluated true).

Progressive disclosure — by default each completed run is a COMPACT summary (scan_run_id, symbol, sweep_params, occurrences, verdict, reliable, and the single strongest horizon under best_horizon). The full per-run blob (all horizons, excursion, diagnostics, warnings) is omitted to stay under the token limit. Pass detail="full" for the complete per-run ScanResult.

Pooled summary — the response always includes pooled_summary, a universe-level meta-analysis combining per-symbol results into one verdict per horizon (Stouffer's weighted Z, Benjamini-Hochberg FDR correction). It reports pooled_verdict, per-horizon stats, included/excluded symbols, and a cross-symbol independence caveat. The pool is computed over the returned set, so filters bias it toward surviving symbols.

Args

• batch_id — from signal_scan
• min_occurrences — move completed results with fewer fires to filtered_out
• significant_only — return only symbols with a horizon at adjusted p-value < 0.05
• detail — "summary" (default) or "full"