diff --git a/.github/workflows/sync-cloud-run-env.yml b/.github/workflows/sync-cloud-run-env.yml
index b33bacf..da471fc 100644
--- a/.github/workflows/sync-cloud-run-env.yml
+++ b/.github/workflows/sync-cloud-run-env.yml
@@ -48,6 +48,7 @@ jobs:
       LONGBRIDGE_FEATURE_SNAPSHOT_MANIFEST_PATH: ${{ vars.LONGBRIDGE_FEATURE_SNAPSHOT_MANIFEST_PATH }}
       LONGBRIDGE_STRATEGY_CONFIG_PATH: ${{ vars.LONGBRIDGE_STRATEGY_CONFIG_PATH }}
       LONGBRIDGE_STRATEGY_PLUGIN_MOUNTS_JSON: ${{ vars.LONGBRIDGE_STRATEGY_PLUGIN_MOUNTS_JSON }}
+      LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD: ${{ vars.LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD }}
       # Optional strategy overrides; leave unset to inherit the UsEquityStrategies profile defaults.
       INCOME_THRESHOLD_USD: ${{ vars.INCOME_THRESHOLD_USD }}
       QQQI_INCOME_RATIO: ${{ vars.QQQI_INCOME_RATIO }}
@@ -337,6 +338,12 @@ jobs:
             remove_env_vars+=("LONGBRIDGE_STRATEGY_PLUGIN_MOUNTS_JSON")
           fi
 
+          if [ -n "${LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD:-}" ]; then
+            env_pairs+=("LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD=${LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD}")
+          else
+            remove_env_vars+=("LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD")
+          fi
+
           if [ -n "${LONGBRIDGE_DRY_RUN_ONLY:-}" ]; then
             env_pairs+=("LONGBRIDGE_DRY_RUN_ONLY=${LONGBRIDGE_DRY_RUN_ONLY}")
           else
diff --git a/README.md b/README.md
index a514ba2..57c8e79 100644
--- a/README.md
+++ b/README.md
@@ -70,6 +70,7 @@ Telegram notifications include structured execution and heartbeat messages, with
 | `LONGBRIDGE_DRY_RUN_ONLY` | No | Set to `true` to keep the selected deployment in dry-run mode. |
 | `LONGBRIDGE_DEBUG_POSITION_SNAPSHOT` | No | Set to `true` to log raw LongBridge position quantity and available quantity for troubleshooting. |
 | `LONGBRIDGE_STRATEGY_PLUGIN_MOUNTS_JSON` | No | Optional LongBridge-side strategy plugin mount JSON. The plugin artifact controls mode; platform config must not set `mode`. |
+| `LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD` | No | Safe-haven/cash-sweep target values below this USD amount are kept as cash instead of buying BOXX/BIL. Default `1000`. |
 | `INCOME_THRESHOLD_USD` | No | Optional strategy override for the `tqqq_growth_income` income-layer threshold. Leave unset to use the strategy package default. |
 | `QQQI_INCOME_RATIO` | No | Optional strategy override for QQQI's share of the `tqqq_growth_income` income layer, 0–1. |
 | `NOTIFY_LANG` | No | Notification language: `en` (English, default) or `zh` (Chinese) |
@@ -119,17 +120,17 @@ Recommended setup:
   - Optional fallback only: `TELEGRAM_TOKEN`
 - **GitHub Environment: `longbridge-paper`**
   - Variables: `CLOUD_RUN_REGION`, `CLOUD_RUN_SERVICE`, `ACCOUNT_PREFIX`, `ACCOUNT_REGION`, `RUNTIME_TARGET_JSON`, `STRATEGY_PROFILE`, `LONGPORT_SECRET_NAME`, `LONGPORT_APP_KEY_SECRET_NAME`, `LONGPORT_APP_SECRET_SECRET_NAME`
-  - Optional variables: `LONGBRIDGE_FEATURE_SNAPSHOT_PATH`, `LONGBRIDGE_FEATURE_SNAPSHOT_MANIFEST_PATH`, `LONGBRIDGE_STRATEGY_CONFIG_PATH`, `LONGBRIDGE_DRY_RUN_ONLY`, `LONGBRIDGE_DEBUG_POSITION_SNAPSHOT`, `INCOME_THRESHOLD_USD`, `QQQI_INCOME_RATIO` (strategy overrides only; leave unset to inherit `UsEquityStrategies`)
+  - Optional variables: `LONGBRIDGE_FEATURE_SNAPSHOT_PATH`, `LONGBRIDGE_FEATURE_SNAPSHOT_MANIFEST_PATH`, `LONGBRIDGE_STRATEGY_CONFIG_PATH`, `LONGBRIDGE_DRY_RUN_ONLY`, `LONGBRIDGE_DEBUG_POSITION_SNAPSHOT`, `LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD`, `INCOME_THRESHOLD_USD`, `QQQI_INCOME_RATIO` (strategy overrides only; leave unset to inherit `UsEquityStrategies`)
   - Current live example: `STRATEGY_PROFILE=mega_cap_leader_rotation_top50_balanced`
   - Recommended secret-name values: `longport-app-key-paper`, `longport-app-secret-paper`
 - **GitHub Environment: `longbridge-sg`**
   - Variables: `CLOUD_RUN_REGION`, `CLOUD_RUN_SERVICE`, `ACCOUNT_PREFIX`, `ACCOUNT_REGION`, `RUNTIME_TARGET_JSON`, `STRATEGY_PROFILE`, `LONGPORT_SECRET_NAME`, `LONGPORT_APP_KEY_SECRET_NAME`, `LONGPORT_APP_SECRET_SECRET_NAME`
-  - Optional variables: `LONGBRIDGE_FEATURE_SNAPSHOT_PATH`, `LONGBRIDGE_FEATURE_SNAPSHOT_MANIFEST_PATH`, `LONGBRIDGE_STRATEGY_CONFIG_PATH`, `LONGBRIDGE_DRY_RUN_ONLY`, `LONGBRIDGE_DEBUG_POSITION_SNAPSHOT`, `INCOME_THRESHOLD_USD`, `QQQI_INCOME_RATIO` (strategy overrides only; leave unset to inherit `UsEquityStrategies`)
+  - Optional variables: `LONGBRIDGE_FEATURE_SNAPSHOT_PATH`, `LONGBRIDGE_FEATURE_SNAPSHOT_MANIFEST_PATH`, `LONGBRIDGE_STRATEGY_CONFIG_PATH`, `LONGBRIDGE_DRY_RUN_ONLY`, `LONGBRIDGE_DEBUG_POSITION_SNAPSHOT`, `LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD`, `INCOME_THRESHOLD_USD`, `QQQI_INCOME_RATIO` (strategy overrides only; leave unset to inherit `UsEquityStrategies`)
   - Current live example: `STRATEGY_PROFILE=soxl_soxx_trend_income`
   - Recommended secret-name values: `longport-app-key-sg`, `longport-app-secret-sg`
 - **GitHub Environment: `longbridge-hk`**
   - Variables: `CLOUD_RUN_REGION`, `CLOUD_RUN_SERVICE`, `ACCOUNT_PREFIX`, `ACCOUNT_REGION`, `RUNTIME_TARGET_JSON`, `STRATEGY_PROFILE`, `LONGPORT_SECRET_NAME`, `LONGPORT_APP_KEY_SECRET_NAME`, `LONGPORT_APP_SECRET_SECRET_NAME`
-  - Optional variables: `LONGBRIDGE_FEATURE_SNAPSHOT_PATH`, `LONGBRIDGE_FEATURE_SNAPSHOT_MANIFEST_PATH`, `LONGBRIDGE_STRATEGY_CONFIG_PATH`, `LONGBRIDGE_DRY_RUN_ONLY`, `LONGBRIDGE_DEBUG_POSITION_SNAPSHOT`, `INCOME_THRESHOLD_USD`, `QQQI_INCOME_RATIO` (strategy overrides only; leave unset to inherit `UsEquityStrategies`)
+  - Optional variables: `LONGBRIDGE_FEATURE_SNAPSHOT_PATH`, `LONGBRIDGE_FEATURE_SNAPSHOT_MANIFEST_PATH`, `LONGBRIDGE_STRATEGY_CONFIG_PATH`, `LONGBRIDGE_DRY_RUN_ONLY`, `LONGBRIDGE_DEBUG_POSITION_SNAPSHOT`, `LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD`, `INCOME_THRESHOLD_USD`, `QQQI_INCOME_RATIO` (strategy overrides only; leave unset to inherit `UsEquityStrategies`)
   - Current live example: `STRATEGY_PROFILE=tech_communication_pullback_enhancement`
   - Recommended secret-name values: `longport-app-key-hk`, `longport-app-secret-hk`
 
@@ -230,6 +231,7 @@ Telegram 通知包含结构化的调仓和心跳消息，支持中英文切换
 | `LONGBRIDGE_DRY_RUN_ONLY` | 否 | 设为 `true` 时，该部署保持 dry-run。 |
 | `LONGBRIDGE_DEBUG_POSITION_SNAPSHOT` | 否 | 设为 `true` 时输出 LongBridge 原始持仓数量和可卖数量，便于排查。 |
 | `LONGBRIDGE_STRATEGY_PLUGIN_MOUNTS_JSON` | 否 | 可选的 LongBridge 侧策略插件挂载 JSON。插件 artifact 自带模式；平台配置不要设置 `mode`。 |
+| `LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD` | 否 | `BOXX`/`BIL` 等避险现金替代标的目标金额低于该 USD 门槛时保留现金，不买入。默认 `1000`。 |
 | `INCOME_THRESHOLD_USD` | 否 | 可选的 `tqqq_growth_income` 收入层启动阈值覆盖（策略 override）。不填时使用策略包默认值。 |
 | `QQQI_INCOME_RATIO` | 否 | 可选的 QQQI 收入层占比覆盖，0–1（策略 override）。 |
 | `NOTIFY_LANG` | 否 | 通知语言: `en`（英文，默认）或 `zh`（中文） |
@@ -279,17 +281,17 @@ Secret Manager 中需存在 `LONGPORT_SECRET_NAME` 指定的密钥（默认: `lo
   - 仅保留为 fallback：`TELEGRAM_TOKEN`
 - **GitHub Environment: `longbridge-paper`**
   - Variables: `CLOUD_RUN_REGION`、`CLOUD_RUN_SERVICE`、`ACCOUNT_PREFIX`、`ACCOUNT_REGION`、`RUNTIME_TARGET_JSON`、`STRATEGY_PROFILE`、`LONGPORT_SECRET_NAME`、`LONGPORT_APP_KEY_SECRET_NAME`、`LONGPORT_APP_SECRET_SECRET_NAME`
-  - 可选 Variables: `LONGBRIDGE_FEATURE_SNAPSHOT_PATH`、`LONGBRIDGE_FEATURE_SNAPSHOT_MANIFEST_PATH`、`LONGBRIDGE_STRATEGY_CONFIG_PATH`、`LONGBRIDGE_DRY_RUN_ONLY`、`LONGBRIDGE_DEBUG_POSITION_SNAPSHOT`、`INCOME_THRESHOLD_USD`、`QQQI_INCOME_RATIO`（仅策略 override；不填则继承 `UsEquityStrategies`）
+  - 可选 Variables: `LONGBRIDGE_FEATURE_SNAPSHOT_PATH`、`LONGBRIDGE_FEATURE_SNAPSHOT_MANIFEST_PATH`、`LONGBRIDGE_STRATEGY_CONFIG_PATH`、`LONGBRIDGE_DRY_RUN_ONLY`、`LONGBRIDGE_DEBUG_POSITION_SNAPSHOT`、`LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD`、`INCOME_THRESHOLD_USD`、`QQQI_INCOME_RATIO`（仅策略 override；不填则继承 `UsEquityStrategies`）
   - 当前线上示例：`STRATEGY_PROFILE=mega_cap_leader_rotation_top50_balanced`
   - 建议的 secret-name 值：`longport-app-key-paper`、`longport-app-secret-paper`
 - **GitHub Environment: `longbridge-sg`**
   - Variables: `CLOUD_RUN_REGION`、`CLOUD_RUN_SERVICE`、`ACCOUNT_PREFIX`、`ACCOUNT_REGION`、`RUNTIME_TARGET_JSON`、`STRATEGY_PROFILE`、`LONGPORT_SECRET_NAME`、`LONGPORT_APP_KEY_SECRET_NAME`、`LONGPORT_APP_SECRET_SECRET_NAME`
-  - 可选 Variables: `LONGBRIDGE_FEATURE_SNAPSHOT_PATH`、`LONGBRIDGE_FEATURE_SNAPSHOT_MANIFEST_PATH`、`LONGBRIDGE_STRATEGY_CONFIG_PATH`、`LONGBRIDGE_DRY_RUN_ONLY`、`LONGBRIDGE_DEBUG_POSITION_SNAPSHOT`、`INCOME_THRESHOLD_USD`、`QQQI_INCOME_RATIO`（仅策略 override；不填则继承 `UsEquityStrategies`）
+  - 可选 Variables: `LONGBRIDGE_FEATURE_SNAPSHOT_PATH`、`LONGBRIDGE_FEATURE_SNAPSHOT_MANIFEST_PATH`、`LONGBRIDGE_STRATEGY_CONFIG_PATH`、`LONGBRIDGE_DRY_RUN_ONLY`、`LONGBRIDGE_DEBUG_POSITION_SNAPSHOT`、`LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD`、`INCOME_THRESHOLD_USD`、`QQQI_INCOME_RATIO`（仅策略 override；不填则继承 `UsEquityStrategies`）
   - 当前线上示例：`STRATEGY_PROFILE=soxl_soxx_trend_income`
   - 建议的 secret-name 值：`longport-app-key-sg`、`longport-app-secret-sg`
 - **GitHub Environment: `longbridge-hk`**
   - Variables: `CLOUD_RUN_REGION`、`CLOUD_RUN_SERVICE`、`ACCOUNT_PREFIX`、`ACCOUNT_REGION`、`RUNTIME_TARGET_JSON`、`STRATEGY_PROFILE`、`LONGPORT_SECRET_NAME`、`LONGPORT_APP_KEY_SECRET_NAME`、`LONGPORT_APP_SECRET_SECRET_NAME`
-  - 可选 Variables: `LONGBRIDGE_FEATURE_SNAPSHOT_PATH`、`LONGBRIDGE_FEATURE_SNAPSHOT_MANIFEST_PATH`、`LONGBRIDGE_STRATEGY_CONFIG_PATH`、`LONGBRIDGE_DRY_RUN_ONLY`、`LONGBRIDGE_DEBUG_POSITION_SNAPSHOT`、`INCOME_THRESHOLD_USD`、`QQQI_INCOME_RATIO`（仅策略 override；不填则继承 `UsEquityStrategies`）
+  - 可选 Variables: `LONGBRIDGE_FEATURE_SNAPSHOT_PATH`、`LONGBRIDGE_FEATURE_SNAPSHOT_MANIFEST_PATH`、`LONGBRIDGE_STRATEGY_CONFIG_PATH`、`LONGBRIDGE_DRY_RUN_ONLY`、`LONGBRIDGE_DEBUG_POSITION_SNAPSHOT`、`LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD`、`INCOME_THRESHOLD_USD`、`QQQI_INCOME_RATIO`（仅策略 override；不填则继承 `UsEquityStrategies`）
   - 当前线上示例：`STRATEGY_PROFILE=tech_communication_pullback_enhancement`
   - 建议的 secret-name 值：`longport-app-key-hk`、`longport-app-secret-hk`
 
diff --git a/application/execution_service.py b/application/execution_service.py
index febfe85..0c089da 100644
--- a/application/execution_service.py
+++ b/application/execution_service.py
@@ -88,10 +88,53 @@ class ExecutionCycleResult:
     action_done: bool
 
 
+DEFAULT_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD = 1000.0
+
+
 def _noop_sleep(_seconds):
     return None
 
 
+def _safe_haven_cash_symbols(*, portfolio: dict, allocation: dict) -> tuple[str, ...]:
+    symbols: list[str] = []
+    for symbol in allocation.get("safe_haven_symbols", ()):
+        normalized = str(symbol or "").strip().upper()
+        if normalized:
+            symbols.append(normalized)
+    cash_sweep_symbol = str(portfolio.get("cash_sweep_symbol") or "").strip().upper()
+    if cash_sweep_symbol:
+        symbols.append(cash_sweep_symbol)
+    return tuple(dict.fromkeys(symbols))
+
+
+def _apply_safe_haven_cash_substitution(
+    *,
+    plan,
+    portfolio,
+    allocation,
+    threshold_usd,
+) -> tuple[dict, dict]:
+    threshold = max(0.0, float(threshold_usd or 0.0))
+    target_values = {
+        str(symbol).strip().upper(): float(value or 0.0)
+        for symbol, value in dict(allocation.get("targets") or {}).items()
+    }
+    if threshold <= 0.0:
+        return dict(plan or {}), {**dict(allocation or {}), "targets": target_values}
+
+    changed = False
+    for symbol in _safe_haven_cash_symbols(portfolio=portfolio, allocation=allocation):
+        target_value = float(target_values.get(symbol, 0.0) or 0.0)
+        if 0.0 < target_value < threshold:
+            target_values[symbol] = 0.0
+            changed = True
+    adjusted_allocation = {**dict(allocation or {}), "targets": target_values}
+    adjusted_plan = dict(plan or {})
+    if changed:
+        adjusted_plan["allocation"] = adjusted_allocation
+    return adjusted_plan, adjusted_allocation
+
+
 def _normalize_cash_by_currency(raw_cash) -> dict[str, float]:
     if not isinstance(raw_cash, Mapping):
         return {}
@@ -249,6 +292,7 @@ def execute_rebalance_cycle(
     post_sell_refresh_attempts=1,
     post_sell_refresh_interval_sec=0.0,
     sleeper=_noop_sleep,
+    safe_haven_cash_substitute_threshold_usd=DEFAULT_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD,
 ) -> ExecutionCycleResult:
     logs: list[str] = []
     skip_logs: list[str] = []
@@ -264,6 +308,12 @@ def execute_rebalance_cycle(
     market_values = dict(portfolio["market_values"])
     quantities = dict(portfolio["quantities"])
     sellable_quantities = dict(portfolio["sellable_quantities"])
+    plan, allocation = _apply_safe_haven_cash_substitution(
+        plan=plan,
+        portfolio=portfolio,
+        allocation=allocation,
+        threshold_usd=safe_haven_cash_substitute_threshold_usd,
+    )
     target_values = dict(allocation["targets"])
     cash_sweep_symbol = str(portfolio.get("cash_sweep_symbol") or "").strip().upper()
     available_cash = float(portfolio["liquid_cash"])
@@ -533,6 +583,12 @@ def record_dry_run(symbol, side, quantity, price, *, order_type):
                     best_refreshed_state = refreshed_state
                     break
             plan, portfolio, execution, allocation = best_refreshed_state
+            plan, allocation = _apply_safe_haven_cash_substitution(
+                plan=plan,
+                portfolio=portfolio,
+                allocation=allocation,
+                threshold_usd=safe_haven_cash_substitute_threshold_usd,
+            )
             threshold_value = float(execution["trade_threshold_value"])
             limit_order_symbols = set(
                 allocation.get("risk_symbols", ()) + allocation.get("income_symbols", ())
@@ -692,7 +748,14 @@ def record_dry_run(symbol, side, quantity, price, *, order_type):
             notify_issue=notify_issue,
         )
         if cash_sweep_price is not None and cash_sweep_price > 0.0 and investable_cash > cash_sweep_price * 2:
-            quantity = int(investable_cash // cash_sweep_price)
+            substitution_threshold = max(
+                0.0,
+                float(safe_haven_cash_substitute_threshold_usd or 0.0),
+            )
+            if substitution_threshold <= 0.0 or investable_cash >= substitution_threshold:
+                quantity = int(investable_cash // cash_sweep_price)
+            else:
+                quantity = 0
             if quantity > 0:
                 quantity_text = format_quantity(quantity)
                 if dry_run_only:
diff --git a/application/rebalance_service.py b/application/rebalance_service.py
index 754dc54..7e6b54e 100644
--- a/application/rebalance_service.py
+++ b/application/rebalance_service.py
@@ -214,6 +214,7 @@ def fetch_replanned_state():
         post_sell_refresh_attempts=config.post_sell_refresh_attempts,
         post_sell_refresh_interval_sec=config.post_sell_refresh_interval_sec,
         sleeper=config.sleeper or _noop_sleep,
+        safe_haven_cash_substitute_threshold_usd=config.safe_haven_cash_substitute_threshold_usd,
     )
     execution = execution_result.execution
     logs = list(execution_result.logs)
diff --git a/application/runtime_composer.py b/application/runtime_composer.py
index 384ea68..05b1c63 100644
--- a/application/runtime_composer.py
+++ b/application/runtime_composer.py
@@ -39,6 +39,7 @@ class LongBridgeRuntimeComposer:
     limit_buy_premium: float
     order_poll_interval_sec: int
     order_poll_max_attempts: int
+    safe_haven_cash_substitute_threshold_usd: float
     dry_run_only: bool = False
     broker_adapters: Any = None
     strategy_adapters: Any = None
@@ -184,6 +185,7 @@ def build_rebalance_config(self, *, strategy_plugin_signals=()) -> LongBridgeReb
             dry_run_only=self.dry_run_only,
             post_sell_refresh_attempts=self.order_poll_max_attempts,
             post_sell_refresh_interval_sec=self.order_poll_interval_sec,
+            safe_haven_cash_substitute_threshold_usd=self.safe_haven_cash_substitute_threshold_usd,
             sleeper=self.sleeper,
             extra_notification_lines=getattr(
                 self.strategy_adapters,
@@ -228,6 +230,7 @@ def build_runtime_composer(
     limit_buy_premium: float,
     order_poll_interval_sec: int,
     order_poll_max_attempts: int,
+    safe_haven_cash_substitute_threshold_usd: float,
     dry_run_only: bool,
     dry_run_only_override: bool | None = None,
     broker_adapters: Any,
@@ -269,6 +272,7 @@ def build_runtime_composer(
         limit_buy_premium=float(limit_buy_premium),
         order_poll_interval_sec=int(order_poll_interval_sec),
         order_poll_max_attempts=int(order_poll_max_attempts),
+        safe_haven_cash_substitute_threshold_usd=float(safe_haven_cash_substitute_threshold_usd),
         dry_run_only=bool(dry_run_only if dry_run_only_override is None else dry_run_only_override),
         broker_adapters=broker_adapters,
         strategy_adapters=strategy_adapters,
diff --git a/application/runtime_dependencies.py b/application/runtime_dependencies.py
index 00faf7f..d3bad10 100644
--- a/application/runtime_dependencies.py
+++ b/application/runtime_dependencies.py
@@ -20,6 +20,7 @@ class LongBridgeRebalanceConfig:
     dry_run_only: bool = False
     post_sell_refresh_attempts: int = 1
     post_sell_refresh_interval_sec: float = 0.0
+    safe_haven_cash_substitute_threshold_usd: float = 1000.0
     sleeper: Callable[[float], None] | None = None
     extra_notification_lines: tuple[str, ...] = ()
 
diff --git a/main.py b/main.py
index d75fa0d..6fb6bc4 100644
--- a/main.py
+++ b/main.py
@@ -92,6 +92,7 @@ def get_project_id():
 
 # Token refresh: days before expiry to trigger refresh
 TOKEN_REFRESH_THRESHOLD_DAYS = 30
+DEFAULT_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD = 1000.0
 
 SEPARATOR = "━━━━━━━━━━━━━━━━━━"
 
@@ -149,6 +150,16 @@ def log_runtime_warning(message):
 )
 
 
+def _safe_haven_cash_substitute_threshold_usd() -> float:
+    return float(
+        getattr(
+            RUNTIME_SETTINGS,
+            "safe_haven_cash_substitute_threshold_usd",
+            DEFAULT_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD,
+        )
+    )
+
+
 def build_composer(*, dry_run_only_override: bool | None = None):
     return build_runtime_composer(
         project_id=PROJECT_ID,
@@ -171,6 +182,7 @@ def build_composer(*, dry_run_only_override: bool | None = None):
         limit_buy_premium=LIMIT_BUY_PREMIUM,
         order_poll_interval_sec=ORDER_POLL_INTERVAL_SEC,
         order_poll_max_attempts=ORDER_POLL_MAX_ATTEMPTS,
+        safe_haven_cash_substitute_threshold_usd=_safe_haven_cash_substitute_threshold_usd(),
         dry_run_only=RUNTIME_SETTINGS.dry_run_only,
         dry_run_only_override=dry_run_only_override,
         broker_adapters=BROKER_ADAPTERS,
diff --git a/runtime_config_support.py b/runtime_config_support.py
index 7cae507..0bfbd53 100644
--- a/runtime_config_support.py
+++ b/runtime_config_support.py
@@ -23,6 +23,7 @@
 
 DEFAULT_ACCOUNT_REGION = "DEFAULT"
 DEFAULT_LONGPORT_SECRET_NAME = "longport_token_hk"
+DEFAULT_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD = 1000.0
 
 
 @dataclass(frozen=True)
@@ -38,6 +39,7 @@ class PlatformRuntimeSettings:
     tg_token: str | None
     tg_chat_id: str | None
     dry_run_only: bool
+    safe_haven_cash_substitute_threshold_usd: float = DEFAULT_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD
     debug_position_snapshot: bool = False
     income_threshold_usd: float | None = None
     qqqi_income_ratio: float | None = None
@@ -77,6 +79,10 @@ def load_platform_runtime_settings(
     project_id_resolver: Callable[[], str | None],
 ) -> PlatformRuntimeSettings:
     account_prefix = os.getenv("ACCOUNT_PREFIX", "DEFAULT")
+    safe_haven_cash_substitute_threshold_usd = resolve_optional_float_env(
+        os.environ,
+        "LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD",
+    )
     runtime_target = resolve_runtime_target_from_env(
         env=os.environ,
         expected_platform_id=LONGBRIDGE_PLATFORM,
@@ -112,6 +118,11 @@ def load_platform_runtime_settings(
         tg_token=os.getenv("TELEGRAM_TOKEN"),
         tg_chat_id=os.getenv("GLOBAL_TELEGRAM_CHAT_ID"),
         dry_run_only=resolve_bool_value(os.getenv("LONGBRIDGE_DRY_RUN_ONLY")),
+        safe_haven_cash_substitute_threshold_usd=(
+            max(0.0, safe_haven_cash_substitute_threshold_usd)
+            if safe_haven_cash_substitute_threshold_usd is not None
+            else DEFAULT_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD
+        ),
         debug_position_snapshot=resolve_bool_value(os.getenv("LONGBRIDGE_DEBUG_POSITION_SNAPSHOT")),
         income_threshold_usd=resolve_optional_float_env(os.environ, "INCOME_THRESHOLD_USD"),
         qqqi_income_ratio=_qqqi_income_ratio_env(),
diff --git a/scripts/print_strategy_switch_env_plan.py b/scripts/print_strategy_switch_env_plan.py
index 61fd952..4f0f520 100644
--- a/scripts/print_strategy_switch_env_plan.py
+++ b/scripts/print_strategy_switch_env_plan.py
@@ -70,7 +70,10 @@ def build_switch_plan(profile: str, *, account_region: str | None = None) -> dic
         "LONGPORT_APP_KEY_SECRET_NAME",
         "LONGPORT_APP_SECRET_SECRET_NAME",
     ]
-    optional_env = ["LONGBRIDGE_DRY_RUN_ONLY"]
+    optional_env = [
+        "LONGBRIDGE_DRY_RUN_ONLY",
+        "LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD",
+    ]
     remove_if_present: list[str] = []
     notes = [
         "Keep ACCOUNT_PREFIX and ACCOUNT_REGION aligned to the current paper or SG service identity.",
diff --git a/tests/test_rebalance_service.py b/tests/test_rebalance_service.py
index 1a69f6e..5c6b46f 100644
--- a/tests/test_rebalance_service.py
+++ b/tests/test_rebalance_service.py
@@ -23,6 +23,7 @@
 sys.modules["requests"] = requests_stub
 try:
     from application import rebalance_service
+    from application.execution_service import execute_rebalance_cycle
     from application.runtime_dependencies import LongBridgeRebalanceConfig, LongBridgeRebalanceRuntime
     from notifications.telegram import build_translator
     from quant_platform_kit.common.models import ExecutionReport, PortfolioSnapshot, Position, QuoteSnapshot
@@ -155,6 +156,61 @@ def _build_snapshot(plan, *, phase=""):
 
 
 class RebalanceServiceNotificationTests(unittest.TestCase):
+    def test_safe_haven_target_below_cash_substitute_threshold_stays_cash(self):
+        submitted_orders = []
+        plan = _build_plan(
+            strategy_symbols=("BOXX",),
+            safe_haven_symbols=("BOXX",),
+            targets={"BOXX": 750.0},
+            market_values={"BOXX": 0.0},
+            sellable_quantities={"BOXX": 0},
+            quantities={"BOXX": 0},
+            current_min_trade=10.0,
+            trade_threshold_value=10.0,
+            investable_cash=750.0,
+            market_status="Cash substitute",
+            deploy_ratio_text="0.0%",
+            income_ratio_text="0.0%",
+            income_locked_ratio_text="0.0%",
+            signal_message="Small safe-haven sleeve",
+            available_cash=750.0,
+            total_strategy_equity=750.0,
+            portfolio_rows=(("BOXX",),),
+        )
+
+        result = execute_rebalance_cycle(
+            trade_context=object(),
+            plan=plan,
+            portfolio=plan["portfolio"],
+            execution=plan["execution"],
+            allocation=plan["allocation"],
+            fetch_replanned_state=lambda: (
+                plan,
+                plan["portfolio"],
+                plan["execution"],
+                plan["allocation"],
+            ),
+            market_data_port=CallableMarketDataPort(
+                quote_loader=lambda symbol: QuoteSnapshot(
+                    symbol=symbol,
+                    as_of="2026-04-21",
+                    last_price=100.0,
+                )
+            ),
+            estimate_max_purchase_quantity=lambda *_args, **_kwargs: 10,
+            execution_port=CallableExecutionPort(lambda order_intent: submitted_orders.append(order_intent)),
+            notify_issue=lambda _title, _detail: None,
+            translator=build_translator("zh"),
+            with_prefix=lambda message: message,
+            limit_sell_discount=0.995,
+            limit_buy_premium=1.0,
+            safe_haven_cash_substitute_threshold_usd=1000.0,
+        )
+
+        self.assertFalse(result.action_done)
+        self.assertEqual(submitted_orders, [])
+        self.assertEqual(result.allocation["targets"]["BOXX"], 0.0)
+
     def test_run_strategy_prefers_portfolio_port_runtime_path(self):
         sent_messages = []
         observed = {}
@@ -549,20 +605,20 @@ def test_strategy_target_rebuys_cash_sweep_symbol_after_buy_skip(self):
             strategy_symbols=("SOXL", "SOXX", "BOXX"),
             risk_symbols=("SOXL", "SOXX"),
             safe_haven_symbols=("BOXX",),
-            targets={"SOXL": 0.0, "SOXX": 163.14, "BOXX": 924.46},
+            targets={"SOXL": 0.0, "SOXX": 163.14, "BOXX": 1224.46},
             market_values={"SOXL": 0.0, "SOXX": 0.0, "BOXX": 699.54},
             sellable_quantities={"SOXL": 0, "SOXX": 0, "BOXX": 6},
             quantities={"SOXL": 0, "SOXX": 0, "BOXX": 6},
             current_min_trade=100.0,
             trade_threshold_value=100.0,
-            investable_cash=891.03,
+            investable_cash=1191.03,
             market_status="🧯 过热降档（SOXX）",
             deploy_ratio_text="15.0%",
             income_ratio_text="0.0%",
             income_locked_ratio_text="0.0%",
             signal_message="SOXX 目标仓位 15.0%",
-            available_cash=923.66,
-            total_strategy_equity=1087.60,
+            available_cash=1223.66,
+            total_strategy_equity=1387.60,
             portfolio_rows=(("SOXL", "SOXX"), ("BOXX",)),
         )
 
@@ -575,10 +631,10 @@ def test_strategy_target_rebuys_cash_sweep_symbol_after_buy_skip(self):
         self.assertIn("🔔 【调仓指令】", sent_messages[0])
         self.assertIn("SOXX.US 目标差额 $163.14", sent_messages[0])
         self.assertIn("SOXX.US 目标差额 $163.14 未超过 1 股价格 $504.60", sent_messages[0])
-        self.assertNotIn("可投资现金 $891.03 不足买入 1 股", sent_messages[0])
+        self.assertNotIn("可投资现金 $1191.03 不足买入 1 股", sent_messages[0])
         self.assertNotIn("市价卖出] BOXX", sent_messages[0])
         self.assertNotIn("市价买入] SOXX", sent_messages[0])
-        self.assertIn("市价买入] BOXX: 7股", sent_messages[0])
+        self.assertIn("市价买入] BOXX: 10股", sent_messages[0])
         self.assertIn("买入说明", sent_messages[0])
         self.assertNotIn("限价买入] SOXX", sent_messages[0])
 
@@ -588,32 +644,32 @@ def test_target_gap_below_one_share_does_not_report_cash_shortage(self):
             risk_symbols=("SOXL", "SOXX"),
             income_symbols=("QQQI", "SPYI"),
             safe_haven_symbols=("BOXX",),
-            targets={"SOXL": 636.28, "SOXX": 218.01, "BOXX": 105.08, "QQQI": 0.0, "SPYI": 0.0},
+            targets={"SOXL": 636.28, "SOXX": 218.01, "BOXX": 1005.08, "QQQI": 0.0, "SPYI": 0.0},
             market_values={"SOXL": 636.28, "SOXX": 218.01, "BOXX": 0.0, "QQQI": 0.0, "SPYI": 0.0},
             sellable_quantities={"SOXL": 4, "SOXX": 0.4326, "BOXX": 0, "QQQI": 0, "SPYI": 0},
             quantities={"SOXL": 4, "SOXX": 0.4326, "BOXX": 0, "QQQI": 0, "SPYI": 0},
             current_min_trade=100.0,
             trade_threshold_value=100.0,
-            investable_cash=164.98,
+            investable_cash=1064.98,
             market_status="🚀 风险开启（SOXX+SOXL）",
             deploy_ratio_text="90.0%",
             income_ratio_text="0.0%",
             income_locked_ratio_text="0.0%",
             signal_message="SOXX 站上 140 日门槛线，持有 SOXL 70.0% + SOXX 20.0%",
-            available_cash=196.50,
-            total_strategy_equity=1050.79,
+            available_cash=1096.50,
+            total_strategy_equity=1950.79,
             portfolio_rows=(("SOXL", "SOXX"), ("BOXX", "QQQI", "SPYI")),
         )
 
         sent_messages, _, _ = self._run_strategy(
             plan,
-            prices={"BOXX.US": 116.74},
+            prices={"BOXX.US": 1167.40},
             dry_run_only=True,
         )
 
         self.assertEqual(len(sent_messages), 1)
         self.assertIn("💓 【心跳检测】", sent_messages[0])
-        self.assertIn("BOXX.US 目标差额 $105.08 未超过 1 股价格 $116.74", sent_messages[0])
+        self.assertIn("BOXX.US 目标差额 $1005.08 未超过 1 股价格 $1167.40", sent_messages[0])
         self.assertNotIn("可投资现金 $164.98 不足买入 1 股", sent_messages[0])
 
     def test_strategy_target_buy_floors_to_cash_backed_whole_shares(self):
@@ -769,20 +825,20 @@ def test_market_buy_floors_to_whole_shares(self):
         plan = _build_plan(
             strategy_symbols=("BOXX",),
             safe_haven_symbols=("BOXX",),
-            targets={"BOXX": 150.0},
+            targets={"BOXX": 1150.0},
             market_values={"BOXX": 0.0},
             sellable_quantities={"BOXX": 0},
             quantities={"BOXX": 0},
             current_min_trade=100.0,
             trade_threshold_value=100.0,
-            investable_cash=200.0,
+            investable_cash=1200.0,
             market_status="🚀 风险开启（BOXX）",
             deploy_ratio_text="100.0%",
             income_ratio_text="0.0%",
             income_locked_ratio_text="0.0%",
             signal_message="BOXX 目标仓位 100.0%",
-            available_cash=200.0,
-            total_strategy_equity=150.0,
+            available_cash=1200.0,
+            total_strategy_equity=1150.0,
             portfolio_rows=(("BOXX",),),
         )
 
@@ -793,8 +849,8 @@ def test_market_buy_floors_to_whole_shares(self):
         )
 
         self.assertEqual(len(sent_messages), 1)
-        self.assertIn("市价买入] BOXX: 1股", sent_messages[0])
-        self.assertNotIn("市价买入] BOXX: 1.5股", sent_messages[0])
+        self.assertIn("市价买入] BOXX: 10股", sent_messages[0])
+        self.assertNotIn("市价买入] BOXX: 10.5股", sent_messages[0])
 
     def test_zero_target_sell_uses_sellable_quantity_not_price_derived_floor(self):
         plan = _build_plan(
@@ -1015,7 +1071,7 @@ def test_cash_sweep_symbol_can_fund_buy_when_investable_cash_is_zero(self):
             quantities={"SOXL": 0, "SOXX": 0, "BOXX": 5},
             current_min_trade=100.0,
             trade_threshold_value=100.0,
-            investable_cash=500.0,
+            investable_cash=1200.0,
             market_status="🧯 过热降档（SOXX）",
             deploy_ratio_text="15.0%",
             income_ratio_text="0.0%",
@@ -1302,14 +1358,14 @@ def test_dry_run_rebuys_cash_sweep_symbol_with_remaining_investable_cash(self):
             quantities={"SOXL": 0, "SOXX": 0, "BOXX": 0},
             current_min_trade=100.0,
             trade_threshold_value=100.0,
-            investable_cash=500.0,
+            investable_cash=1200.0,
             market_status="🧯 过热降档（SOXX）",
             deploy_ratio_text="0.0%",
             income_ratio_text="0.0%",
             income_locked_ratio_text="0.0%",
             signal_message="无其他买单，仅保留现金回补",
-            available_cash=500.0,
-            total_strategy_equity=500.0,
+            available_cash=1200.0,
+            total_strategy_equity=1200.0,
             portfolio_rows=(("SOXL", "SOXX"), ("BOXX",)),
         )
 
diff --git a/tests/test_runtime_composer.py b/tests/test_runtime_composer.py
index f3552eb..4e279d6 100644
--- a/tests/test_runtime_composer.py
+++ b/tests/test_runtime_composer.py
@@ -51,6 +51,7 @@ def fake_bootstrap_builder(**kwargs):
         limit_buy_premium=1.005,
         order_poll_interval_sec=1,
         order_poll_max_attempts=8,
+        safe_haven_cash_substitute_threshold_usd=1000.0,
         dry_run_only=True,
         runtime_target=build_runtime_target(
             platform_id="longbridge",
@@ -129,3 +130,4 @@ def fake_bootstrap_builder(**kwargs):
     assert config.limit_buy_premium == 1.005
     assert config.strategy_display_name == "SOXL/SOXX 半导体趋势收益"
     assert config.dry_run_only is True
+    assert config.safe_haven_cash_substitute_threshold_usd == 1000.0
diff --git a/tests/test_runtime_config_support.py b/tests/test_runtime_config_support.py
index 85645ad..9bb8130 100644
--- a/tests/test_runtime_config_support.py
+++ b/tests/test_runtime_config_support.py
@@ -20,6 +20,7 @@
 from runtime_config_support import (
     DEFAULT_ACCOUNT_REGION,
     DEFAULT_LONGPORT_SECRET_NAME,
+    DEFAULT_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD,
     infer_account_region,
     load_platform_runtime_settings,
 )
@@ -99,6 +100,10 @@ def test_load_platform_runtime_settings_uses_defaults_with_explicit_strategy_pro
         self.assertIsNone(settings.tg_token)
         self.assertIsNone(settings.tg_chat_id)
         self.assertFalse(settings.dry_run_only)
+        self.assertEqual(
+            settings.safe_haven_cash_substitute_threshold_usd,
+            DEFAULT_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD,
+        )
         self.assertFalse(settings.debug_position_snapshot)
         self.assertIsNotNone(settings.runtime_target)
         self.assertEqual(settings.runtime_target.platform_id, "longbridge")
@@ -176,6 +181,19 @@ def test_debug_position_snapshot_is_loaded_from_env(self):
 
         self.assertTrue(settings.debug_position_snapshot)
 
+    def test_safe_haven_cash_substitute_threshold_is_loaded_from_env(self):
+        with patch.dict(
+            os.environ,
+            {
+                "RUNTIME_TARGET_JSON": runtime_target_json(SAMPLE_STRATEGY_PROFILE),
+                "LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD": "750",
+            },
+            clear=True,
+        ):
+            settings = load_platform_runtime_settings(project_id_resolver=lambda: "project-1")
+
+        self.assertEqual(settings.safe_haven_cash_substitute_threshold_usd, 750.0)
+
     def test_strategy_plugin_mounts_are_loaded_from_env(self):
         mount_config = '{"strategy_plugins":[{"strategy":"soxl_soxx_trend_income","plugin":"crisis_response_shadow","signal_path":"gs://bucket/latest_signal.json"}]}'
         with patch.dict(
@@ -463,6 +481,7 @@ def test_print_strategy_switch_env_plan_for_global_etf_rotation(self):
         self.assertEqual(plan["input_mode"], "market_history")
         self.assertFalse(plan["requires_snapshot_artifacts"])
         self.assertFalse(plan["requires_strategy_config_path"])
+        self.assertIn("LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD", plan["optional_env"])
         self.assertIn("LONGBRIDGE_FEATURE_SNAPSHOT_PATH", plan["remove_if_present"])
 
     def test_print_strategy_switch_env_plan_for_russell(self):
diff --git a/tests/test_sync_cloud_run_env_workflow.sh b/tests/test_sync_cloud_run_env_workflow.sh
index 0d2182b..4e98d09 100644
--- a/tests/test_sync_cloud_run_env_workflow.sh
+++ b/tests/test_sync_cloud_run_env_workflow.sh
@@ -44,6 +44,7 @@ grep -Fq 'LONGBRIDGE_FEATURE_SNAPSHOT_PATH: ${{ vars.LONGBRIDGE_FEATURE_SNAPSHOT
 grep -Fq 'LONGBRIDGE_FEATURE_SNAPSHOT_MANIFEST_PATH: ${{ vars.LONGBRIDGE_FEATURE_SNAPSHOT_MANIFEST_PATH }}' "$workflow_file"
 grep -Fq 'LONGBRIDGE_STRATEGY_CONFIG_PATH: ${{ vars.LONGBRIDGE_STRATEGY_CONFIG_PATH }}' "$workflow_file"
 grep -Fq 'LONGBRIDGE_STRATEGY_PLUGIN_MOUNTS_JSON: ${{ vars.LONGBRIDGE_STRATEGY_PLUGIN_MOUNTS_JSON }}' "$workflow_file"
+grep -Fq 'LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD: ${{ vars.LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD }}' "$workflow_file"
 grep -Fq 'INCOME_THRESHOLD_USD: ${{ vars.INCOME_THRESHOLD_USD }}' "$workflow_file"
 grep -Fq 'QQQI_INCOME_RATIO: ${{ vars.QQQI_INCOME_RATIO }}' "$workflow_file"
 grep -Fq 'LONGBRIDGE_DRY_RUN_ONLY: ${{ vars.LONGBRIDGE_DRY_RUN_ONLY }}' "$workflow_file"
@@ -79,6 +80,8 @@ grep -Fq 'LONGBRIDGE_FEATURE_SNAPSHOT_PATH=${LONGBRIDGE_FEATURE_SNAPSHOT_PATH}'
 grep -Fq 'LONGBRIDGE_FEATURE_SNAPSHOT_MANIFEST_PATH=${LONGBRIDGE_FEATURE_SNAPSHOT_MANIFEST_PATH}' "$workflow_file"
 grep -Fq 'LONGBRIDGE_STRATEGY_CONFIG_PATH=${LONGBRIDGE_STRATEGY_CONFIG_PATH}' "$workflow_file"
 grep -Fq 'LONGBRIDGE_STRATEGY_PLUGIN_MOUNTS_JSON=${LONGBRIDGE_STRATEGY_PLUGIN_MOUNTS_JSON}' "$workflow_file"
+grep -Fq 'LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD=${LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD}' "$workflow_file"
+grep -Fq 'remove_env_vars+=("LONGBRIDGE_SAFE_HAVEN_CASH_SUBSTITUTE_THRESHOLD_USD")' "$workflow_file"
 grep -Fq 'LONGBRIDGE_DRY_RUN_ONLY=${LONGBRIDGE_DRY_RUN_ONLY}' "$workflow_file"
 grep -Fq 'INCOME_THRESHOLD_USD=${INCOME_THRESHOLD_USD}' "$workflow_file"
 grep -Fq 'QQQI_INCOME_RATIO=${QQQI_INCOME_RATIO}' "$workflow_file"