dCLV close leg: read priced captures only (the unfixed twin)

computeDirectionalForRow selected the latest closing_captures row with no
missed_reason filter, while its sibling attachClosingProb has had one since the
CLV instrument repair. closing_captures records a refusal for every prop x book
x side on every cycle after first pitch, so a refusal always carries a later
captured_at than the last real price - latest-first returned a refusal on
26,448 of 26,448 identity groups, and computeDirectionalClv refuses on a
missedReason. That is why dclv_state has been 'unknown' on 100% of rows since
Session 64.

Measured on 600 real settled rows: 100% unknown becomes flat 45.8%, negative
23.0%, positive 22.8%, unknown 8.3%. ClvBadge will render MOVED TOWARD US 114
and MOVED AWAY 117 per 600 - near-symmetric, which is the honest shape.

Existing rows do not recompute (first-computation-wins). The re-stamp is
described in BUILD-STATE, not run.

The grade is untouched. No rows deleted.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

src/services/ledgerService.js

@@ -605,10 +605,25 @@ async function computeDirectionalForRow(sb, sport, row, deps = {}) {
         .eq('sport', sport).eq('player_key', row.player_key).eq('stat', row.stat)
         .eq('side', row.side).eq('game_date', row.game_date)
         .order('captured_at', { ascending: true }).limit(1),
+      // ── FIX B1 — READ ONLY PRICED CAPTURES ──────────────────────────
+      // This select had no `missed_reason` filter while its sibling
+      // `attachClosingProb` (above) has had one since the CLV instrument repair.
+      // `closing_captures` records a refusal for every prop x book x side on
+      // every ~20-minute cycle AFTER first pitch, so those rows always carry a
+      // LATER `captured_at` than the last real price — and `latest-first`
+      // therefore returned a refusal on 100% of identity groups (measured:
+      // 26,448 of 26,448 over five days). `computeDirectionalClv` refuses on a
+      // `missedReason`, which is why `dclv_state` has been 'unknown' on every
+      // row since Session 64.
+      //
+      // The filter belongs BEFORE the limit: latest of the PRICED rows, not
+      // "the latest row, hopefully priced".
       sb.from('closing_captures')
         .select('over_odds, under_odds, missed_reason, captured_at')
         .eq('sport', sport).eq('player_key', row.player_key).eq('stat', row.stat)
         .eq('side', row.side).eq('game_date', row.game_date)
+        .is('missed_reason', null)
+        .not('over_odds', 'is', null).not('under_odds', 'is', null)
         .order('captured_at', { ascending: false }).limit(1),
     ]);
     const lock = snaps && snaps[0];
@@ -1085,5 +1100,9 @@ module.exports = {
     rowsFromSnapshot, computeClv, clvResultOf, indexProps, gameIdFor,
     dateET, sideOf, oddsForSide, isConfigured, CONFLICT,
     teamOpponentFor, teamsMatch, clvBucketIndex, CLV_BUCKETS,
+    // B1 — exported so the close leg is directly testable. It was untestable,
+    // and that is why the missing missed_reason filter survived while its
+    // sibling was fixed a few lines above it.
+    computeDirectionalForRow,
   },
 };