diff --git a/deploy/profile-scripts/per_stock_reassess.py b/deploy/profile-scripts/per_stock_reassess.py
index 31f1e687..d4747d8a 100644
--- a/deploy/profile-scripts/per_stock_reassess.py
+++ b/deploy/profile-scripts/per_stock_reassess.py
@@ -66,18 +66,10 @@ def _build_full_analysis(code, entry, result):
         eh = result.get("entry_high")
     sl = result.get("stop_loss") or entry.get("stop_loss", 0)
     tp = result.get("take_profit") or entry.get("take_profit", 0)
-    # 2026-08-18 修复 rr_ratio：统一用 entry/stop/tp 算（与 promote 同口径 rr = (tp-mid)/(mid-sl)）
-    # strategy_lifecycle 的 rr_ratio 可能算错（6.5 vs 实际 8.64），统一用 entry/stop/tp 算最准
-    _el_rr = result.get("entry_low") if result.get("entry_low") is not None else entry.get("entry_low", 0)
-    _eh_rr = result.get("entry_high") if result.get("entry_high") is not None else entry.get("entry_high", 0)
-    _sl_rr = result.get("stop_loss") or entry.get("stop_loss", 0)
-    _tp_rr = result.get("take_profit") or entry.get("take_profit", 0)
-    if _el_rr > 0 and _eh_rr > _el_rr and _sl_rr > 0 and _tp_rr > 0:
-        _mid_rr = (_el_rr + _eh_rr) / 2
-        rr = round((_tp_rr - _mid_rr) / (_mid_rr - _sl_rr), 2) if _mid_rr > _sl_rr else 0
-    else:
-        # 无有效区间/止损/止盈时，用 holding 的旧 rr_ratio 或 0
-        rr = entry.get("rr_ratio", 0)
+    # 2026-08-18 修复：删除 rr_ratio 覆盖——RR 由 scanner 定（candidates.rr=holding.rr_ratio），
+    # per_stock_reassess 不该重算/覆盖（它该用 holding 的 rr_ratio，不重算）。
+    # 正确架构：RR 是 scanner 定义的单一事实来源，per_stock_reassess 只读不重算。
+    rr = entry.get("rr_ratio", 0)
     act = result.get("action", "")
     
     # ── 从DB拉取大盘、基本面、资金流 ──
diff --git a/deploy/profile-scripts/s2_panic_v2_scanner.py b/deploy/profile-scripts/s2_panic_v2_scanner.py
index 2783f75f..052bce2b 100644
--- a/deploy/profile-scripts/s2_panic_v2_scanner.py
+++ b/deploy/profile-scripts/s2_panic_v2_scanner.py
@@ -159,14 +159,17 @@ def main():
     inserted = 0
     for code, name, sig in hits:
         reasons = (f"rsi={sig['rsi']} mcap_q={sig['mcap_q']} score={sig['score']}")
+        # 2026-08-18 补 rr 列（断链 bug：promote 的 _rr is None 全跳过，与 b_td1_v3 同口径）
+        _mid_v = (sig['price'] * 0.98 + sig['price']) / 2
+        _rr_v = round((sig["target"] - _mid_v) / (_mid_v - sig["stop_loss"]), 2) if _mid_v > sig["stop_loss"] > 0 else 0
         conn.execute(
-            "INSERT INTO candidates (code, name, sector, reason, entry_range, stop_loss, target, source_strategy, created_at) "
-            "VALUES (?,?,?,?,?,?,?,?,datetime('now','localtime')) "
+            "INSERT INTO candidates (code, name, sector, reason, entry_range, stop_loss, target, rr, source_strategy, created_at) "
+            "VALUES (?,?,?,?,?,?,?,?,?,datetime('now','localtime')) "
             "ON CONFLICT(code) DO UPDATE SET "
             "name=excluded.name, sector=excluded.sector, reason=excluded.reason, "
-            "entry_range=excluded.entry_range, stop_loss=excluded.stop_loss, target=excluded.target, source_strategy=excluded.source_strategy",
+            "entry_range=excluded.entry_range, stop_loss=excluded.stop_loss, target=excluded.target, rr=excluded.rr, source_strategy=excluded.source_strategy",
             (code, code, "s2_panic_v2", reasons,
-             f"{sig['price']*0.98:.2f}~{sig['price']:.2f}", sig["stop_loss"], sig["target"], "s2_panic_v2"))
+             f"{sig['price']*0.98:.2f}~{sig['price']:.2f}", sig["stop_loss"], sig["target"], _rr_v, "s2_panic_v2"))
         inserted += 1
         print(f"  🟢 {code} {name} 价{sig['price']} score={sig['score']} {reasons}", flush=True)
     conn.commit()
diff --git a/deploy/profile-scripts/strategy_lifecycle.py b/deploy/profile-scripts/strategy_lifecycle.py
index 8e1d0e45..4237f5e4 100644
--- a/deploy/profile-scripts/strategy_lifecycle.py
+++ b/deploy/profile-scripts/strategy_lifecycle.py
@@ -1179,15 +1179,11 @@ def reassess_strategy(code, name, price, cost, shares, current_action,
                     found = True
                     break
 
-    # ----- 风险回报比最终计算 -----
-    # 2026-08-18 修复：买入价用买入区中值(entry_mid)算 risk，不是现价
-    # 现价不是买入价——实盘按 entry_mid 买入，risk=entry_mid-stop，reward=tp-entry_mid
-    _el = result.get("entry_low") or 0
-    _eh = result.get("entry_high") or 0
-    _entry_mid = (_el + _eh) / 2 if (_el > 0 and _eh > _el) else price
-    risk = max(_entry_mid - new_stop, _entry_mid * 0.01)
-    reward = max(new_target - _entry_mid, 0)
-    rr_ratio = reward / risk if risk > 0 else 0
+    # ----- 风险回报比（不重算，RR 由 scanner 定）-----
+    # 2026-08-18 修复：RR 由 scanner 定（candidates.rr=holding.rr_ratio），
+    # strategy_lifecycle 管持仓状态（信号/止损移动），不重算/重写 RR。
+    # 正确架构：RR 是 scanner 定义的单一事实来源，lifecycle 只读不重算。
+    rr_ratio = result.get("rr_ratio") or 0
 
     # ----- 状态判断 -----
     if is_deep_loss: