CLV instrument repair: fix attachClosingProb read + recoverable market_unavailable

The closing_prob funnel collapsed 100k priced captures -> 59 usable. Root cause
(VERIFIED against prod, join key is PERFECT with 0 mismatches):
- attachClosingProb read closing_captures with .limit(50000) and NO ORDER BY on
  a 730k-row table that is 86% refusal rows -> saw ~7% for MLB, missed most
  priced closes and declared 200+ rows closeless that HAD a capture.
- market_unavailable_reason was write-once/terminal, so a row wrongly declared
  (truncated read / premature declaration before the capture was visible) could
  never recover even once its genuine capture existed. 298 rows (204 MLB + 94
  WNBA) were stuck this way.

Fix (CLV computation only — no grade/locked_odds/outcome touched):
- Read ONLY priced captures (missed_reason IS NULL, both odds NOT NULL), scoped
  to the candidate rows' game_dates -> small AND complete, no arbitrary truncation.
- Drop the market_unavailable exclusion from candidates; make it a re-checkable
  absence: a genuine close now UPGRADES the row (writes closing_prob, clears the
  verdict). closing_prob stays write-once (first true close wins). No capture +
  past game -> still declared absent (honest). No churn on already-absent rows.
- New internal trigger POST /api/internal/ledger/attach-closing[/:sport] for
  backfill + verification (scheduler already runs attach per tick).

Recovers ~312 usable closes (59 -> ~371), MLB included. Capture itself was
healthy all along (94.9% MLB / 95.8% WNBA per-prop coverage). Full suite 3835
green (17/17 instrument tests incl. 2 new recovery cases), web build exit 0.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VsztNChZ7vEvSR61AuMhD1

src/routes/internal.js

@@ -405,6 +405,33 @@ router.post('/ledger/settle', async (req, res) => {
   }
 });
 
+/**
+ * POST /api/internal/ledger/attach-closing[/:sport] — CLV instrument repair.
+ * Attach the de-vigged closing probability from `closing_captures` onto locked
+ * model-record rows. Recovers rows a truncated read / premature verdict wrongly
+ * declared closeless (market_unavailable is a re-checkable absence, not terminal).
+ * Idempotent: `closing_prob` is write-once; a row with a genuine close never
+ * changes, and a row with no capture stays honestly absent.
+ */
+async function attachClosingHandler(req, res) {
+  const ledger = require('../services/ledgerService');
+  const ALL = ['mlb', 'wnba', 'nba', 'soccer'];
+  const only = String(req.params.sport || '').toLowerCase();
+  const sports = only ? [only] : ALL;
+  try {
+    const results = {};
+    for (const sp of sports) results[sp] = await ledger.attachClosingProb(sp, { limit: Number(req.query.limit) || undefined });
+    return res.json({ ok: true, results });
+  } catch (err) {
+    const message = err && err.message ? err.message : String(err);
+    console.error('[internal/ledger/attach-closing] failed:', message);
+    return res.status(500).json({ ok: false, error: message });
+  }
+}
+// Express 5 (path-to-regexp v8) has no `/:sport?` optional param — register both.
+router.post('/ledger/attach-closing', attachClosingHandler);
+router.post('/ledger/attach-closing/:sport', attachClosingHandler);
+
 /**
  * POST /api/internal/newsletter/send (Session S7, a1) — assemble today's
  * VYNDR REPORT from the pipeline (snapshot signals + streak lens + the

src/services/ledgerService.js

@@ -342,11 +342,17 @@ async function attachClosingProb(sport, opts = {}) {
   const sb = opts.sb || defaultClient();
   const sp = String(sport || '').toLowerCase();
 
+  // Candidates: any locked model-record row still lacking a real close. We do
+  // NOT exclude rows already stamped market_unavailable_reason — that verdict is
+  // a RE-CHECKABLE absence, not terminal. A capture we couldn't see before (the
+  // read was truncated) or that landed after a premature declaration must be
+  // able to upgrade the row. `closing_prob` itself stays write-once (the filter
+  // below), so the FIRST true close still wins and is never rewritten.
   const { data: rows, error } = await sb.from('ledger_entries')
-    .select('id, player_key, stat, side, game_date')
+    .select('id, player_key, stat, side, game_date, market_unavailable_reason')
     .is('user_id', null).eq('sport', sp)
-    .is('closing_prob', null).is('market_unavailable_reason', null)
+    .is('closing_prob', null)
-    .limit(opts.limit || 2000);
+    .limit(opts.limit || 5000);
   if (error) return { updated: 0, error: error.message };
   if (!rows || !rows.length) return { updated: 0, absent: 0 };
 
@@ -355,10 +361,20 @@ async function attachClosingProb(sport, opts = {}) {
   // the grade-time fair price uses, which is what makes lock and close
   // comparable at all. Asking this table for a `fair_prob` column is a bug: it
   // has none, and every row then looks closeless.
+  //
+  // READ FIX (CLV instrument repair): the table is ~86% refusal rows, and the
+  // old `.limit(50000)` with no ORDER BY read an arbitrary slice — for MLB
+  // (730k rows) it saw ~7% and declared 200+ rows closeless that HAD a priced
+  // capture. Read only PRICED captures, scoped to the candidate rows' game
+  // dates, so the set is small AND complete.
   const { devigTwoWay } = require('../utils/devig');
-  const { data: caps } = await sb.from('closing_captures')
+  const dates = [...new Set(rows.map((r) => r.game_date).filter(Boolean))];
+  let capsQuery = sb.from('closing_captures')
     .select('player_key, stat, side, game_date, over_odds, under_odds, captured_at, missed_reason')
-    .eq('sport', sp).limit(50000);
+    .eq('sport', sp).is('missed_reason', null)
+    .not('over_odds', 'is', null).not('under_odds', 'is', null);
+  if (dates.length) capsQuery = capsQuery.in('game_date', dates);
+  const { data: caps } = await capsQuery.limit(opts.capLimit || 200000);
 
   // Latest usable capture per identity = the TRUE close.
   const best = new Map();
@@ -378,17 +394,28 @@ async function attachClosingProb(sport, opts = {}) {
   // rows market-unavailable would be premature absence — as dishonest in the
   // other direction as imputing one. Only a past game can be declared closeless.
   const cutoff = opts.beforeDate || todayET();
+  let recovered = 0;
   for (const r of rows) {
     const hit = best.get(`${r.player_key}|${r.stat}|${r.side}|${r.game_date}`);
-    if (!hit && String(r.game_date) >= String(cutoff)) continue; // still capturable
+    if (hit) {
-    const patch = hit
+      // A real close — write it and CLEAR any prior (premature/truncation-bug)
-      ? { closing_prob: hit.fair_prob, closing_captured_at: hit.captured_at }
+      // market-unavailable verdict. This is the recovery path: a row wrongly
-      : { market_unavailable_reason: 'no_usable_close' };
+      // declared closeless is upgraded the moment its genuine capture is seen.
-    const { error: e } = await sb.from('ledger_entries').update(patch).eq('id', r.id);
+      const patch = { closing_prob: hit.fair_prob, closing_captured_at: hit.captured_at, market_unavailable_reason: null };
+      const { error: e } = await sb.from('ledger_entries').update(patch).eq('id', r.id);
+      if (e) continue;
+      updated += 1;
+      if (r.market_unavailable_reason) recovered += 1;
+      continue;
+    }
+    // No usable close found for this row.
+    if (String(r.game_date) >= String(cutoff)) continue;   // future/today — a close can still arrive
+    if (r.market_unavailable_reason) continue;              // already honestly declared absent — leave it
+    const { error: e } = await sb.from('ledger_entries').update({ market_unavailable_reason: 'no_usable_close' }).eq('id', r.id);
     if (e) continue;
-    if (hit) updated += 1; else absent += 1;
+    absent += 1;
   }
-  return { updated, absent, candidates: rows.length };
+  return { updated, absent, recovered, candidates: rows.length };
 }
 
 /**

tests/unit/projectionInstrument.test.js

@@ -78,6 +78,7 @@ describe('attachClosingProb — the market half', () => {
       from: (t) => ({
         select: () => ({
           is: function () { return this; }, eq: function () { return this; },
+          not: function () { return this; }, in: function () { return this; },
           limit: async () => ({ data: t === 'ledger_entries' ? rows : caps, error: null }),
         }),
         update: (patch) => ({ eq: async (_c, id) => { updates.push({ id, patch }); return { error: null }; } }),
@@ -150,6 +151,30 @@ describe('attachClosingProb — the market half', () => {
     const fn = src.slice(src.indexOf('async function attachClosingProb'));
     expect(fn).toMatch(/\.is\('closing_prob', null\)/);
   });
+
+  // CLV instrument repair — the write-once verdict on market_unavailable_reason
+  // was permanent, so a row wrongly declared closeless (truncated read /
+  // premature declaration) could never recover even though its capture existed.
+  it('RECOVERS a row previously declared market_unavailable when a real close now exists', async () => {
+    const sb = makeSb({
+      rows: [{ ...ROW, market_unavailable_reason: 'no_usable_close' }],
+      caps: [{ player_key: 'josh bell', stat: 'hits', side: 'over', game_date: '2026-07-20', over_odds: -210, under_odds: 170, captured_at: '2026-07-20T22:50:00Z' }],
+    });
+    const out = await ledger.attachClosingProb('mlb', { sb, beforeDate: '2026-07-21' });
+    expect(out.updated).toBe(1);
+    expect(out.recovered).toBe(1);
+    expect(sb.updates[0].patch.closing_prob).toBe(0.647);
+    // the buggy verdict is CLEARED, not left stale beside a real close.
+    expect(sb.updates[0].patch.market_unavailable_reason).toBeNull();
+  });
+
+  it('leaves an already-declared row untouched when there is STILL no capture (no churn, no re-write)', async () => {
+    const sb = makeSb({ rows: [{ ...ROW, market_unavailable_reason: 'no_usable_close' }], caps: [] });
+    const out = await ledger.attachClosingProb('mlb', { sb, beforeDate: '2026-07-21' });
+    expect(out.updated).toBe(0);
+    expect(out.absent).toBe(0);
+    expect(sb.updates).toHaveLength(0); // no write at all
+  });
 });
 
 describe('what stays measurable when the market is absent', () => {