From 7c8fe2fe9f0f29548c838f75c9cba47f9302390c Mon Sep 17 00:00:00 2001
From: Yijia-Xiao <yijia-xiao@outlook.com>
Date: Sat, 13 Jun 2026 20:39:52 +0000
Subject: [PATCH] fix(data): normalize symbols on the identity and reflection
 paths

resolve_instrument_identity and the reflection return lookup queried Yahoo with
the raw ticker, so broker/forex/commodity symbols (XAUUSD, BTCUSD, EURUSD)
failed identity or could mismatch the priced instrument even though the price
path already normalized them. Route both through normalize_symbol (#983, #984).
---
 tests/test_symbol_normalization_paths.py  | 54 +++++++++++++++++++++++
 tradingagents/agents/utils/agent_utils.py |  7 ++-
 tradingagents/graph/trading_graph.py      |  7 ++-
 3 files changed, 66 insertions(+), 2 deletions(-)
 create mode 100644 tests/test_symbol_normalization_paths.py

diff --git a/tests/test_symbol_normalization_paths.py b/tests/test_symbol_normalization_paths.py
new file mode 100644
index 000000000..a023d24a6
--- /dev/null
+++ b/tests/test_symbol_normalization_paths.py
@@ -0,0 +1,54 @@
+"""Symbol normalization must apply on every yfinance path, not just price fetch.
+
+Regression tests for #983 (instrument identity) and #984 (reflection returns):
+a broker symbol like XAUUSD must resolve to the same Yahoo symbol (GC=F) that
+the price path uses, so identity and realized-return lookups hit the right
+instrument instead of failing/mismatching.
+"""
+import pandas as pd
+
+import tradingagents.agents.utils.agent_utils as au
+import tradingagents.graph.trading_graph as tg
+from tradingagents.graph.trading_graph import TradingAgentsGraph
+
+
+def test_identity_lookup_normalizes_symbol(monkeypatch):
+    seen = {}
+
+    class FakeTicker:
+        def __init__(self, symbol):
+            seen["symbol"] = symbol
+
+        @property
+        def info(self):
+            return {"longName": "Gold Futures", "quoteType": "FUTURE"}
+
+    monkeypatch.setattr(au.yf, "Ticker", FakeTicker)
+    au.resolve_instrument_identity.cache_clear()
+
+    identity = au.resolve_instrument_identity("XAUUSD")
+
+    assert seen["symbol"] == "GC=F"  # normalized, not the raw broker symbol
+    assert identity.get("company_name") == "Gold Futures"
+
+
+def test_fetch_returns_normalizes_symbol(monkeypatch):
+    queried = []
+
+    class FakeTicker:
+        def __init__(self, symbol):
+            queried.append(symbol)
+
+        def history(self, *args, **kwargs):
+            return pd.DataFrame({"Close": [100.0, 101.0, 102.0, 103.0, 104.0, 105.0, 106.0]})
+
+    monkeypatch.setattr(tg.yf, "Ticker", FakeTicker)
+
+    # _fetch_returns does not use ``self``; call unbound to avoid building the graph.
+    raw, alpha, days = TradingAgentsGraph._fetch_returns(
+        None, "XAUUSD", "2025-01-02", holding_days=5, benchmark="SPY"
+    )
+
+    assert queried[0] == "GC=F"  # stock symbol normalized (#984)
+    assert queried[1] == "SPY"   # benchmark left as the canonical symbol
+    assert raw is not None and days is not None
diff --git a/tradingagents/agents/utils/agent_utils.py b/tradingagents/agents/utils/agent_utils.py
index 7476654c0..92113e5e5 100644
--- a/tradingagents/agents/utils/agent_utils.py
+++ b/tradingagents/agents/utils/agent_utils.py
@@ -70,9 +70,14 @@ def resolve_instrument_identity(ticker: str) -> dict:
     recognise the ticker, we return ``{}`` and the caller falls back to
     ticker-only context rather than failing before analysis starts. Cached so
     the lookup happens at most once per ticker per process.
+
+    The symbol is normalized first (e.g. ``XAUUSD`` -> ``GC=F``) so identity
+    resolves for the same instrument the price path actually fetches (#983).
     """
+    from tradingagents.dataflows.symbol_utils import normalize_symbol
+
     try:
-        info = yf.Ticker(ticker.upper()).info or {}
+        info = yf.Ticker(normalize_symbol(ticker)).info or {}
     except Exception as exc:  # noqa: BLE001 — fail open, never block the run
         logger.debug("Could not resolve instrument identity for %s: %s", ticker, exc)
         return {}
diff --git a/tradingagents/graph/trading_graph.py b/tradingagents/graph/trading_graph.py
index 2a851b344..a3de660f7 100644
--- a/tradingagents/graph/trading_graph.py
+++ b/tradingagents/graph/trading_graph.py
@@ -232,12 +232,17 @@ class TradingAgentsGraph:
         actual_holding_days)`` or ``(None, None, None)`` if price data is
         unavailable (too recent, delisted, or network error).
         """
+        from tradingagents.dataflows.symbol_utils import normalize_symbol
+
         try:
             start = datetime.strptime(trade_date, "%Y-%m-%d")
             end = start + timedelta(days=holding_days + 7)  # buffer for weekends/holidays
             end_str = end.strftime("%Y-%m-%d")
 
-            stock = yf.Ticker(ticker).history(start=trade_date, end=end_str)
+            # Normalize so the realized-return lookup hits the same instrument
+            # the analysis priced (e.g. XAUUSD -> GC=F) (#984). The benchmark is
+            # already a canonical Yahoo symbol from ``_resolve_benchmark``.
+            stock = yf.Ticker(normalize_symbol(ticker)).history(start=trade_date, end=end_str)
             bench = yf.Ticker(benchmark).history(start=trade_date, end=end_str)
 
             if len(stock) < 2 or len(bench) < 2: