From 2942655f703b45a67026e740684ab5a7b5410022 Mon Sep 17 00:00:00 2001
From: Yijia-Xiao <yijia-xiao@outlook.com>
Date: Tue, 15 Sep 2026 00:04:25 +0000
Subject: [PATCH] fix(dataflows): map Shanghai .SH tickers to Yahoo's .SS
 (#1260)

- normalize_symbol maps 600519.SH to 600519.SS
- the alpha benchmark is resolved from the normalized symbol
---
 tests/test_memory_log.py                | 2 ++
 tests/test_symbol_utils.py              | 4 ++++
 tradingagents/dataflows/symbol_utils.py | 7 ++++++-
 tradingagents/graph/trading_graph.py    | 4 +++-
 4 files changed, 15 insertions(+), 2 deletions(-)

diff --git a/tests/test_memory_log.py b/tests/test_memory_log.py
index 77629fb4d..28a943498 100644
--- a/tests/test_memory_log.py
+++ b/tests/test_memory_log.py
@@ -598,6 +598,8 @@ class TestDeferredReflection:
                              "benchmark_map": DEFAULT_CONFIG["benchmark_map"]}
         assert TradingAgentsGraph._resolve_benchmark(mock_graph, "600519.SS") == "000001.SS"
         assert TradingAgentsGraph._resolve_benchmark(mock_graph, "000001.SZ") == "399001.SZ"
+        # .SH is the exchange's own suffix; Yahoo spells Shanghai .SS (#1260)
+        assert TradingAgentsGraph._resolve_benchmark(mock_graph, "600519.SH") == "000001.SS"
 
     def test_resolve_benchmark_us_ticker_defaults_to_spy(self):
         """US tickers (no dotted suffix) take the empty-suffix entry."""
diff --git a/tests/test_symbol_utils.py b/tests/test_symbol_utils.py
index f87ca393a..8ddf877c8 100644
--- a/tests/test_symbol_utils.py
+++ b/tests/test_symbol_utils.py
@@ -69,6 +69,10 @@ class TestNormalizeSymbol(unittest.TestCase):
     def test_hk_code_case_insensitive_suffix(self):
         self.assertEqual(normalize_symbol("09992.hk"), "9992.HK")
 
+    def test_shanghai_sh_suffix_maps_to_yahoo_ss(self):
+        self.assertEqual(normalize_symbol("600519.sh"), "600519.SS")
+        self.assertEqual(normalize_symbol("600519.SS"), "600519.SS")
+
 
 @pytest.mark.unit
 class TestNoMarketDataError(unittest.TestCase):
diff --git a/tradingagents/dataflows/symbol_utils.py b/tradingagents/dataflows/symbol_utils.py
index 274cc0889..647f50444 100644
--- a/tradingagents/dataflows/symbol_utils.py
+++ b/tradingagents/dataflows/symbol_utils.py
@@ -11,6 +11,7 @@ differ from the broker / TradingView / MT5 style symbols users often type:
     BTCUSD            BTC-USD           crypto pairs use a ``-`` separator
     SPX500, US500     ^GSPC             index CFDs map to Yahoo index symbols
     09992.HK, 700.HK  9992.HK, 0700.HK  HK codes are zero-padded to 4 digits
+    600519.SH         600519.SS         Yahoo spells Shanghai ``.SS``
 
 Passing the raw broker symbol to Yahoo returns an empty result, which the
 agents previously received as free text and could hallucinate a price
@@ -75,6 +76,7 @@ _YAHOO_SAFE = re.compile(r"^[A-Za-z0-9._\-\^=]+$")
 
 # HKEX codes as Yahoo spells them: the number zero-padded to 4 digits (#957).
 _HK_CODE = re.compile(r"^(\d{1,5})\.HK$")
+_SHANGHAI_SH = re.compile(r"^(\d{6})\.SH$")
 
 
 # Crypto quote currencies that all map to Yahoo's USD pair. Yahoo lists only
@@ -115,7 +117,8 @@ def normalize_symbol(raw: str) -> str:
       3. Forex rule: six letters that are two ISO currency codes -> ``PAIR=X``.
       4. HK rule: a numeric ``.HK`` code -> Yahoo's 4-digit padding
          (``09992.HK`` -> ``9992.HK``, ``700.HK`` -> ``0700.HK``).
-      5. Otherwise the upper-cased symbol is returned unchanged (plain
+      5. Shanghai rule: ``600519.SH`` -> ``600519.SS``.
+      6. Otherwise the upper-cased symbol is returned unchanged (plain
          equities, ETFs, Yahoo-native symbols like ``GC=F`` or ``^GSPC``).
 
     A trailing ``+`` (broker CFD marker, e.g. ``XAUUSD+``) is stripped before
@@ -138,6 +141,8 @@ def normalize_symbol(raw: str) -> str:
         canonical = f"{s}=X"
     elif hk := _HK_CODE.match(s):
         canonical = f"{int(hk.group(1)):04d}.HK"
+    elif sh := _SHANGHAI_SH.match(s):
+        canonical = f"{sh.group(1)}.SS"
     else:
         canonical = s
 
diff --git a/tradingagents/graph/trading_graph.py b/tradingagents/graph/trading_graph.py
index 4d732644b..900266535 100644
--- a/tradingagents/graph/trading_graph.py
+++ b/tradingagents/graph/trading_graph.py
@@ -274,11 +274,13 @@ class TradingAgentsGraph:
         entry, which is the right default because the alpha calculation works
         in USD.
         """
+        from tradingagents.dataflows.symbol_utils import normalize_symbol
+
         explicit = self.config.get("benchmark_ticker")
         if explicit:
             return explicit
         benchmark_map = self.config.get("benchmark_map", {})
-        ticker_upper = ticker.upper()
+        ticker_upper = normalize_symbol(ticker)
         for suffix, benchmark in benchmark_map.items():
             if suffix and ticker_upper.endswith(suffix.upper()):
                 return benchmark