fix: verify squeezes on a median volume baseline and ingest from the premarket

KITT on 2026-09-25 (+557%, 9.3M shares, the day's biggest mover) was benched
as "gap 557% squeeze unverified (rvol 1.54)". Two separate faults:

1. The rvol baseline was the 20-day MEAN, and two 11-14M-share days on
   09-01/09-02 put it at 1.66M against a 333k median. Squeeze verification
   (gap >= 150% needs rvol >= 5) now takes the stronger of the mean-baseline
   reading and the median-baseline one. Both are cumulative volume over a
   full day's baseline, not prorated, so they share the units the rvol >= 5
   boundary was calibrated in. A reverse-split fake still reads ~0.01 on
   both. This applies in the watcher pre-filter, the trigger book's squeeze
   exception and the fast-path population gate. The ranking and the tod-rvol
   arm signal are unchanged.

2. A name's first ingest reached back only to ~09:15 (09:30 + 15 min, or
   the 20-minute refresh floor). KITT was found at 09:35, so its stored day
   began at 09:13 and missed the 09:00 squeeze and ~6M premarket shares.
   The first pass now covers the session from 04:00 ET.

With both fixed, KITT's volume at 09:35 is ~20x its median day.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

research/ingestion/signals.py

@@ -32,6 +32,12 @@ class GapperCandidate:
     # 29.4M). None when no baseline. Legacy relative_volume is kept untouched —
     # validated boundaries (gap-squeeze rvol>=5) are calibrated in its units.
     relative_volume_tod: float | None = None
+    # Squeeze-verification rvol: cum volume vs MEDIAN(20d), NOT prorated — the
+    # legacy units the gap-squeeze rvol>=5 boundary was calibrated in, but on a
+    # baseline one or two prior spike days can't poison. KITT 2026-09-25 (+557%,
+    # 9.3M shares) read legacy 1.54 because two 11-14M days put the 20d mean at
+    # 1.66M vs a 333k median, and was benched as "squeeze unverified".
+    relative_volume_med: float | None = None
 
 
 def _session_elapsed_fraction(now: datetime | None = None) -> float:
@@ -111,6 +117,8 @@ def scan_gappers(
         rvol = (float(cum_volume) / float(adv)) if adv and adv > 0 else 0.0
         rvol_tod = (float(cum_volume) / (float(med_vol) * frac)
                     if med_vol and float(med_vol) > 0 else None)
+        rvol_med = (float(cum_volume) / float(med_vol)
+                    if med_vol and float(med_vol) > 0 else None)
         # either rvol flavour clears the bar: legacy is ~0.1-0.3 early in the
         # session even for a heavy name (cum-so-far vs full-day mean), so the
         # tod variant keeps real early movers from being silently dropped
@@ -127,6 +135,7 @@ def scan_gappers(
                     relative_volume=rvol,
                     vwap=float(vwap) if vwap is not None else None,
                     relative_volume_tod=rvol_tod,
+                    relative_volume_med=rvol_med,
                 )
             )
     candidates.sort(key=lambda c: (c.gap_pct * max(c.relative_volume, 0.1)), reverse=True)

research/ingestion/watcher_task.py

@@ -71,6 +71,9 @@ class ResearchWatchlistProvider:
                     avg_daily_volume=research_query.query_avg_daily_volume(
                         self.con, symbol, session_date
                     ),
+                    median_daily_volume=research_query.query_median_daily_volume(
+                        self.con, symbol, session_date
+                    ),
                     source="manual",
                 )
             )
@@ -89,6 +92,9 @@ class ResearchWatchlistProvider:
                     avg_daily_volume=research_query.query_avg_daily_volume(
                         self.con, symbol, session_date
                     ),
+                    median_daily_volume=research_query.query_median_daily_volume(
+                        self.con, symbol, session_date
+                    ),
                     source="research",
                 )
             )
@@ -124,7 +130,7 @@ class LiveWatchlistProvider:
         for symbol in self.symbols:
             daily = self._daily_bars(symbol)
             prev_close = None
-            adv = None
+            adv = med = None
             if daily:
                 # last bar strictly before the session date
                 prior = [
@@ -135,11 +141,15 @@ class LiveWatchlistProvider:
                     prev_close = float(prior[-1]["c"])
                     recent = prior[-20:]
                     adv = sum(float(b["v"]) for b in recent) / len(recent)
+                    vols = sorted(float(b["v"]) for b in recent)
+                    mid = len(vols) // 2
+                    med = vols[mid] if len(vols) % 2 else (vols[mid - 1] + vols[mid]) / 2
             candidates.append(
                 WatchCandidate(
                     symbol=symbol,
                     previous_close=prev_close,
                     avg_daily_volume=adv,
+                    median_daily_volume=med,
                     source="live",
                 )
             )