perf: parallelize compute_signals with ThreadPoolExecutor

Strategies perform independent network I/O (yfinance calls via _data
helpers), so running them concurrently in a thread pool reduces wall-clock
time from O(n) to O(n/workers).

Addresses code review feedback on sequential strategy computation.

tradingagents/strategies/registry.py

@@ -5,12 +5,15 @@ from __future__ import annotations
 import importlib
 import logging
 import pkgutil
+from concurrent.futures import ThreadPoolExecutor, as_completed
 from typing import Any
 
 from .base import BaseStrategy, Role, StrategySignal
 
 logger = logging.getLogger(__name__)
 
+_MAX_WORKERS = 4  # cap threads; strategies do network I/O, not CPU work
+
 _registry: list[BaseStrategy] = []
 
 
@@ -49,19 +52,32 @@ def reset_registry() -> None:
     _registry.clear()
 
 
+def _run_strategy(
+    strategy: BaseStrategy, ticker: str, date: str, context: dict[str, Any] | None,
+) -> StrategySignal | None:
+    """Execute a single strategy, returning None on failure."""
+    try:
+        return strategy.compute(ticker, date, context)
+    except Exception:
+        logger.warning("Strategy %s failed for %s@%s", strategy.name, ticker, date, exc_info=True)
+        return None
+
+
 def compute_signals(
     ticker: str, date: str, context: dict[str, Any] | None = None
 ) -> list[StrategySignal]:
-    """Run every registered strategy and collect non-None signals."""
+    """Run every registered strategy in parallel and collect non-None signals."""
     _discover()
     signals: list[StrategySignal] = []
-    for strategy in _registry:
-        try:
-            sig = strategy.compute(ticker, date, context)
+    with ThreadPoolExecutor(max_workers=min(_MAX_WORKERS, len(_registry) or 1)) as pool:
+        futures = {
+            pool.submit(_run_strategy, s, ticker, date, context): s
+            for s in _registry
+        }
+        for fut in as_completed(futures):
+            sig = fut.result()
             if sig is not None:
                 signals.append(sig)
-        except Exception:
-            logger.warning("Strategy %s failed for %s@%s", strategy.name, ticker, date, exc_info=True)
     return signals