[RULE-REVIEW] Entries live again: stock and options entries order again (human directive 2026-11-04) (v1.0.1272)

ci/counter_baseline.txt

@@ -35,7 +35,7 @@ conflicting_const        3        non-increasing
 undeclared_py_import     2        non-increasing
 dead_workflow_env        0        non-increasing
 uncapped_surface         0        non-increasing
-assertions               27582    non-decreasing
+assertions               27600    non-decreasing
 harness_rules_checked    84       non-decreasing
 baseline_divergence      0        non-increasing
 dup_precise_literal      3        non-increasing

package.json

@@ -1,6 +1,6 @@
 {
   "name": "rest-express",
-  "version": "1.0.1271",
+  "version": "1.0.1272",
   "type": "module",
   "license": "MIT",
   "scripts": {

research/experiments.md

@@ -113420,3 +113420,24 @@ guessed, not read. Next session after a >24 h uptime window: read /api/health me
 baseline, then ship A8 with the measured number. PRIORITY CHECK: nothing in priority 1–2 is broken; no
 matured experiment is due. MERGE NOTE: docs-only; no deploy impact either way, merge after 4:00 PM ET by routine.
 STARVED: no (A8 is queued but measurement-gated, not capacity-gated).
+
+## 2026-11-04 [RULE-REVIEW] — T-BOT (system_config.py SHADOW_ENTRY_SLEEVES; mirrors server/entryShadow.ts, server/brain.ts) — ENTRIES LIVE AGAIN: stock and options entries order again (human directive)
+VERSION: v1.0.1272 (shipped 2026-11-04 by the parent session; main was v1.0.1271)
+
+Human directive (2026-11-04, verbatim): "figure out the option bleed and i didnt want you to stop trading thing like the floor it a paper account and see what work the best you know the goal is th beat the sp"
+
+Change: SHADOW_ENTRY_SLEEVES "stocks,options" -> "off". The prior value was set 2026-10-28 (v1.0.1210, recorded as a human directive) and was in force 6 sessions. Mirrors updated: server/entryShadow.ts ENTRY_SHADOW_DEFAULTS, server/brain.ts BRAIN_CONFIG_MIRROR. The mechanism is untouched: VOLTRADE_SHADOW_ENTRIES still wins whenever present, so the rollback needs no deploy (set it to "stocks,options", or "options" to re-shadow only the options sleeve).
+
+Downstream chain:
+1. Options opens resume (tier SELL_CSP/BUY_PUT, scanner setups, evaluate_and_execute CSPs/covered calls/spreads, convexity QQQ puts, roll re-opens) -> option buying power is used again -> the options sleeve produces live results again, now with #1582 (a multi-leg strategy is closed as a unit) and #1583 (POS-KILL never kills one leg of a defined-risk structure) live on fresh entries for the first time.
+2. Stock/ETF entries resume -> but in NEUTRAL the floor targets SPY_FLOOR_NEUTRAL 0.90 + DEFENSIVE_FLOOR_NEUTRAL 0.08 = 98% of equity, above the tier engine's 95% invested cap (TIER-KILL "Portfolio invested 96.7% >= 95%" at 14:46:59Z today) -> most stock entries stay blocked while NEUTRAL lasts; BULL (0.70 + 0.08) leaves about 17% of room, CAUTION (0.35 + 0.15) about 45%. This crowd-out is pre-existing and NOT changed here; it is a separate rule question that needs its own evidence.
+
+Prior (stated before any live result): options expectancy per closed lot over the first 30 lots opened on this version lands between -10% and +5% (60% confidence), better than the historical -17.1% because the 11-02 fixes target the same-day churn that carried most of the loss (research/options_exit_study_2026-11-02.md: lots closed within a day lost -$9.8k); stock entries stay rare (< 5 a week) while NEUTRAL persists, because of the floor crowd-out.
+
+Review date: 2026-12-04 (+20 trading days). The human decides keep / re-shadow per sleeve from live expectancy by sleeve (AI Engine page) and the account vs SPY over the same window. Rollback trigger (any one): options expectancy below -15% per closed lot after >= 30 closed lots opened on this version; the drawdown ladder halts (kill switches unchanged); the options-bleed diagnosis (in progress this session) finds a defect that makes new opens lose by construction.
+
+Backtest: not run. This reverts a 6-session-old human-directed entry switch to the pre-2026-10-28 behaviour the live record already covers, and the honest test of the 11-02 fixes is live fills (REASONING STANDARD 9). Logged as an exploratory experiment with the review date above (PROMOTION RULES 3).
+
+Tests: test_entry_shadow.py pins the new default "off" and that the env rollback still shadows both sleeves; server/entryShadow.test.ts the same; server/brain.test.ts keeps every assertion of the old all-shadowed scenario under an explicit config and adds a test for the live default; the flagRaw pin moves to "off". Touched-area suites: Python 89/89, Node/client 48/48.
+
+STARVED: no (options-bleed diagnosis running; the fold-cache race fix is queued).

server/brain.test.ts

@@ -58,7 +58,7 @@ test("long-term evidence: every quoted figure is still in its source file (doc d
 test("flagRaw: env PRESENT wins (even empty / off); absent -> config default; True -> '1'", () => {
   assert.deepEqual(flagRaw("VOLTRADE_REGIME_ENGINE", {}, BRAIN_CONFIG_MIRROR), { value: "1", source: "config" });
   assert.deepEqual(flagRaw("VOLTRADE_REGIME_ENGINE", { VOLTRADE_REGIME_ENGINE: "" }, BRAIN_CONFIG_MIRROR), { value: "", source: "env" });
-  assert.deepEqual(flagRaw("VOLTRADE_SHADOW_ENTRIES", {}, BRAIN_CONFIG_MIRROR), { value: "stocks,options", source: "config" });
+  assert.deepEqual(flagRaw("VOLTRADE_SHADOW_ENTRIES", {}, BRAIN_CONFIG_MIRROR), { value: "off", source: "config" });
   assert.deepEqual(flagRaw("VOLTRADE_REGIME_LABELS", {}, { REGIME_LABELS_DEFAULT: false }), { value: "", source: "config" });
 });
 
@@ -70,31 +70,50 @@ test("parse: selection/sizing sets and the options crisis+vrp rule mirror regime
   assert.deepEqual([...parseOptionsGates("0")], []);
 });
 
-test("status: defaults (both entry sleeves shadowed) — score inputs are SHADOW, floor/exits LIVE, dead code OFF", () => {
+test("status: the 2026-10-28 config (both entry sleeves shadowed) — score inputs are SHADOW, floor/exits LIVE, dead code OFF", () => {
+  // The config default until the 2026-11-04 directive; still reachable via the env rollback.
   const env = {};
-  assert.equal(status("stock_entries", env).status, "shadow");
+  const cfg = { ...BRAIN_CONFIG_MIRROR, SHADOW_ENTRY_SLEEVES: "stocks,options" };
+  assert.equal(status("stock_entries", env, cfg).status, "shadow");
+  assert.equal(status("stock_entries", env, cfg).status_source, "config");
+  assert.equal(status("options_entries", env, cfg).status, "shadow");
+  assert.equal(status("news", env, cfg).status, "shadow");
+  assert.match(status("news", env, cfg).detail, /nothing it touches is ordered/);
+  assert.equal(status("floor_basket", env, cfg).status, "live");
+  assert.equal(status("floor_basket", env, cfg).status_source, "config");
+  assert.equal(status("live_exits", env, cfg).status, "live");
+  assert.equal(status("live_exits", env, cfg).status_source, "code");
+  assert.equal(status("strategy_weights", env, cfg).status, "off");
+  assert.equal(status("google_trends", env, cfg).status, "off");
+  assert.equal(status("ml_v3", env, cfg).status, "shadow");
+  assert.equal(status("forecast", env, cfg).status, "shadow");
+  assert.equal(status("regime_sizing", env, cfg).status, "shadow");
+  // regime guards are gated but stock entries are shadowed -> they only block would-be entries
+  assert.equal(status("sel_knife", env, cfg).status, "shadow");
+  assert.match(status("sel_knife", env, cfg).detail, /would-be entries only/);
+  // the stress multiplier is replaced by the regime engine's sizing on the stock sleeve
+  assert.equal(status("stress_index", env, cfg).status, "off");
+  // convexity: CONVEXITY_OVERLAY.enabled is never read; the options shadow decides
+  assert.equal(status("convexity", env, cfg).status, "shadow");
+  assert.match(status("convexity", env, cfg).detail, /never read/);
+  assert.equal(status("third_leg", env, cfg).status, "shadow");
+});
+
+test("status: defaults since the 2026-11-04 directive (entries live) — entry sleeves and score inputs LIVE from config", () => {
+  const env = {};
+  assert.equal(status("stock_entries", env).status, "live");
   assert.equal(status("stock_entries", env).status_source, "config");
-  assert.equal(status("options_entries", env).status, "shadow");
-  assert.equal(status("news", env).status, "shadow");
-  assert.match(status("news", env).detail, /nothing it touches is ordered/);
+  assert.equal(status("options_entries", env).status, "live");
+  assert.equal(status("options_entries", env).status_source, "config");
+  assert.equal(status("news", env).status, "live");
+  assert.equal(status("sel_knife", env).status, "live");
+  assert.equal(status("csp_block_guards", env).status, "live");
+  assert.equal(status("convexity", env).status, "live");
+  assert.equal(status("third_leg", env).status, "live");
   assert.equal(status("floor_basket", env).status, "live");
-  assert.equal(status("floor_basket", env).status_source, "config");
   assert.equal(status("live_exits", env).status, "live");
-  assert.equal(status("live_exits", env).status_source, "code");
   assert.equal(status("strategy_weights", env).status, "off");
   assert.equal(status("google_trends", env).status, "off");
-  assert.equal(status("ml_v3", env).status, "shadow");
-  assert.equal(status("forecast", env).status, "shadow");
-  assert.equal(status("regime_sizing", env).status, "shadow");
-  // regime guards are gated but stock entries are shadowed -> they only block would-be entries
-  assert.equal(status("sel_knife", env).status, "shadow");
-  assert.match(status("sel_knife", env).detail, /would-be entries only/);
-  // the stress multiplier is replaced by the regime engine's sizing on the stock sleeve
-  assert.equal(status("stress_index", env).status, "off");
-  // convexity: CONVEXITY_OVERLAY.enabled is never read; the options shadow decides
-  assert.equal(status("convexity", env).status, "shadow");
-  assert.match(status("convexity", env).detail, /never read/);
-  assert.equal(status("third_leg", env).status, "shadow");
 });
 
 test("status: env flips — entries live, regime flags off, an explicit empty env value wins over the config default", () => {

server/brain.ts

@@ -47,7 +47,7 @@ export const PRIMARY_BUCKET_OUTCOME = "xa";
 
 /** BASE_CONFIG values the status rules read — a MIRROR used only when the Python call failed (test_brain_evidence.py pins it). */
 export const BRAIN_CONFIG_MIRROR: Readonly<Record<string, unknown>> = Object.freeze({
-  SHADOW_ENTRY_SLEEVES: "stocks,options",
+  SHADOW_ENTRY_SLEEVES: "off",
   REGIME_ENGINE_ENABLED: true,
   REGIME_SELECTION_DEFAULT: "all",
   REGIME_OPTIONS_DEFAULT: "all",

server/entryShadow.test.ts

@@ -27,11 +27,18 @@ function slice(fromMarker: string, toMarker: string): string {
 }
 
 // ── flag ─────────────────────────────────────────────────────────────────────
-test("default (env absent) shadows stocks and options, source config", () => {
-  assert.equal(ENTRY_SHADOW_DEFAULTS.VOLTRADE_SHADOW_ENTRIES, "stocks,options");
-  assert.deepEqual(entryShadowStatus({}), { stocks: true, options: true, source: "config" });
-  assert.equal(isEntryShadowed("stocks", {}), true);
-  assert.equal(isEntryShadowed("options", {}), true);
+test("default (env absent) is live for both sleeves since the 2026-11-04 directive, source config", () => {
+  assert.equal(ENTRY_SHADOW_DEFAULTS.VOLTRADE_SHADOW_ENTRIES, "off");
+  assert.deepEqual(entryShadowStatus({}), { stocks: false, options: false, source: "config" });
+  assert.equal(isEntryShadowed("stocks", {}), false);
+  assert.equal(isEntryShadowed("options", {}), false);
+});
+
+test("the env rollback (VOLTRADE_SHADOW_ENTRIES=stocks,options) still shadows both sleeves", () => {
+  const env = { VOLTRADE_SHADOW_ENTRIES: "stocks,options" };
+  assert.deepEqual(entryShadowStatus(env), { stocks: true, options: true, source: "env" });
+  assert.equal(isEntryShadowed("stocks", env), true);
+  assert.equal(isEntryShadowed("options", env), true);
 });
 
 test("a PRESENT env var wins — every off value turns both sleeves live", () => {

server/entryShadow.ts

@@ -21,8 +21,9 @@
  */
 
 export const ENTRY_SHADOW_ENV = "VOLTRADE_SHADOW_ENTRIES";
+// 2026-11-04 human directive: entries live again ("off"); prior "stocks,options".
 export const ENTRY_SHADOW_DEFAULTS: Readonly<Record<string, string>> = Object.freeze({
-  VOLTRADE_SHADOW_ENTRIES: "stocks,options",
+  VOLTRADE_SHADOW_ENTRIES: "off",
 });
 
 /** shadow_portfolio.SHADOW_ENTRY_DECISION / SHADOW_REPEAT_DECISION. */

system_config.py

@@ -454,7 +454,18 @@
     # (entry_shadow.flag_raw; Node mirror server/entryShadow.ts).
     # Prior behaviour: every entry live (no key). Expected effect: no new
     # stock/ETF/option positions; buying power stays in cash + floor.
-    "SHADOW_ENTRY_SLEEVES": "stocks,options",
+    # 2026-11-04 HUMAN DIRECTIVE (paper account; verbatim in research/
+    # experiments.md this date): the human did not want trading stopped:
+    # trade the sleeves live, see what works best, the goal is to beat SPY.
+    # Entries are LIVE again for both sleeves. Prior value "stocks,options"
+    # (2026-10-28 -> 2026-11-04). Expected effect: stock/ETF entries and
+    # options opens order again (the pre-2026-10-28 behaviour, now with the
+    # 2026-11-02 multi-leg close-as-a-unit and POS-KILL structure-hold fixes
+    # live for the first time); in NEUTRAL the floor (SPY_FLOOR_NEUTRAL 0.90
+    # + DEFENSIVE_FLOOR_NEUTRAL 0.08) still leaves little room under the 95%
+    # exposure cap for new stock entries (unchanged here). Rollback without
+    # a deploy: env VOLTRADE_SHADOW_ENTRIES=stocks,options (or =options).
+    "SHADOW_ENTRY_SLEEVES": "off",
     # Target universe size — top N scored candidates per cycle
     "T1_UNIVERSE_MAX_SIZE": 200,
     # REGIME-SPECIFIC STOP SCALARS 2026-04-22

test_entry_shadow.py

@@ -4,7 +4,9 @@
 exits live." — entry_shadow.py + server/entryShadow.ts.
 
 Pinned here:
-  1. flag: BASE_CONFIG["SHADOW_ENTRY_SLEEVES"] = "stocks,options"; the env
+  1. flag: BASE_CONFIG["SHADOW_ENTRY_SLEEVES"] = "off" since the 2026-11-04
+     human directive (entries live again; "stocks,options" 2026-10-28 ->
+     2026-11-04); the env
      var VOLTRADE_SHADOW_ENTRIES wins whenever it is PRESENT ("off"/"0"/""/
      "none" = nothing shadowed); the Node mirror's default and decision
      constants equal the Python ones.
@@ -73,13 +75,22 @@ def env(self, values):
 
 # ── 1. flag ──────────────────────────────────────────────────────────────────
 class TestFlag(unittest.TestCase):
-    def test_base_config_default_shadows_stocks_and_options(self):
-        self.assertEqual(BASE_CONFIG["SHADOW_ENTRY_SLEEVES"], "stocks,options")
-        self.assertEqual(entry_shadow.shadowed_sleeves({}), {"stocks", "options"})
-        self.assertTrue(entry_shadow.is_shadowed("stocks", {}))
-        self.assertTrue(entry_shadow.is_shadowed("options", {}))
+    def test_base_config_default_is_live_since_the_2026_11_04_directive(self):
+        # 2026-11-04 human directive (paper account): entries live again.
+        # The 2026-10-28 -> 2026-11-04 default was "stocks,options".
+        self.assertEqual(BASE_CONFIG["SHADOW_ENTRY_SLEEVES"], "off")
+        self.assertEqual(entry_shadow.shadowed_sleeves({}), set())
+        self.assertFalse(entry_shadow.is_shadowed("stocks", {}))
+        self.assertFalse(entry_shadow.is_shadowed("options", {}))
         self.assertEqual(entry_shadow.status({}),
-                         {"stocks": True, "options": True, "source": "config", "value": "stocks,options"})
+                         {"stocks": False, "options": False, "source": "config", "value": "off"})
+
+    def test_the_rollback_value_still_shadows_both_sleeves(self):
+        # The env rollback the config comment names must keep working.
+        env = {"VOLTRADE_SHADOW_ENTRIES": "stocks,options"}
+        self.assertEqual(entry_shadow.shadowed_sleeves(env), {"stocks", "options"})
+        self.assertEqual(entry_shadow.status(env),
+                         {"stocks": True, "options": True, "source": "env", "value": "stocks,options"})
 
     def test_present_env_wins_including_off_values(self):
         for v in OFF_VALUES: