diff --git a/sanguo_orchestrator/portfolio_worker.py b/sanguo_orchestrator/portfolio_worker.py
index c6b46b2..0427136 100644
--- a/sanguo_orchestrator/portfolio_worker.py
+++ b/sanguo_orchestrator/portfolio_worker.py
@@ -21,7 +21,7 @@ logger = logging.getLogger(__name__)
 _VPS_HOST = "49.232.102.198"
 _VPS_WORKDIR = r"C:\\sanguo_vnpy_v2"
 _VPS_PYTHON = "python"
-_TIMEOUT = 3600  # 1 hour hard cap (was 600s sync in routes_portfolio)
+_TIMEOUT = 10800  # 3h hard cap (was 3600/1h; 全市场全周期 60-100min/策略不够)
 
 
 def run_portfolio_task(spec: dict) -> Any:
@@ -56,10 +56,21 @@ def run_portfolio_task(spec: dict) -> Any:
     argv, cwd = _build_argv(start, end, cash, benchmark, max_pool, provider_config, strategy)
     logger.info("[portfolio_worker] task=%s running: %s", task_id, " ".join(argv[3:]))
 
-    proc = subprocess.run(
-        argv, cwd=cwd, capture_output=True, text=True,
-        timeout=_TIMEOUT, check=False,
-    )
+    try:
+        proc = subprocess.run(
+            argv, cwd=cwd, capture_output=True, text=True,
+            timeout=_TIMEOUT, check=False,
+        )
+    except subprocess.TimeoutExpired:
+        # 全市场全周期慢→3h仍超→显式log(否则_wait_future静默task.fail难排查)
+        logger.error(
+            "[portfolio_worker] task=%s TIMEOUT after %ss (argv tail=%s)",
+            task_id, _TIMEOUT, " ".join(argv[3:]),
+        )
+        raise RuntimeError(
+            f"runner_backtest timeout after {_TIMEOUT}s "
+            f"(全市场全周期?考虑缩小 universe 或周期)"
+        )
 
     if proc.returncode != 0:
         stderr_tail = (proc.stderr or "")[-2000:]