diff --git a/tests/test_data_stream.py b/tests/test_data_stream.py
index 935c3f2..6f4e4ac 100644
--- a/tests/test_data_stream.py
+++ b/tests/test_data_stream.py
@@ -23,6 +23,7 @@ def test_multi_symbol_stream_get_dataframe_returns_none_when_empty():
 
 def test_multi_symbol_stream_get_dataframe_returns_df_when_full():
     import pandas as pd
+    from src.data_stream import _MIN_CANDLES_FOR_SIGNAL
     stream = MultiSymbolStream(
         symbols=["XRPUSDT", "BTCUSDT", "ETHUSDT"],
         interval="1m",
@@ -32,13 +33,13 @@ def test_multi_symbol_stream_get_dataframe_returns_df_when_full():
         "timestamp": 1000, "open": 1.0, "high": 1.1,
         "low": 0.9, "close": 1.05, "volume": 100.0, "is_closed": True,
     }
-    for i in range(50):
+    for i in range(_MIN_CANDLES_FOR_SIGNAL):
         c = candle.copy()
         c["timestamp"] = 1000 + i
         stream.buffers["xrpusdt"].append(c)
     df = stream.get_dataframe("XRPUSDT")
     assert df is not None
-    assert len(df) == 50
+    assert len(df) == _MIN_CANDLES_FOR_SIGNAL
 
 
 @pytest.mark.asyncio
diff --git a/tests/test_ml_features.py b/tests/test_ml_features.py
index b7a645f..1f3ffff 100644
--- a/tests/test_ml_features.py
+++ b/tests/test_ml_features.py
@@ -49,9 +49,9 @@ def test_build_features_rs_zero_when_btc_ret_zero():
     features = build_features(xrp_df, "LONG", btc_df=btc_df, eth_df=eth_df)
     assert features["xrp_btc_rs"] == 0.0
 
-def test_feature_cols_has_21_items():
+def test_feature_cols_has_23_items():
     from src.ml_features import FEATURE_COLS
-    assert len(FEATURE_COLS) == 21
+    assert len(FEATURE_COLS) == 23
 
 
 def make_df(n=100):
diff --git a/tests/test_ml_filter.py b/tests/test_ml_filter.py
index 27580f3..0c837ae 100644
--- a/tests/test_ml_filter.py
+++ b/tests/test_ml_filter.py
@@ -12,13 +12,19 @@ def make_features(side="LONG") -> pd.Series:
 
 
 def test_no_model_file_is_not_loaded(tmp_path):
-    f = MLFilter(model_path=str(tmp_path / "nonexistent.pkl"))
+    f = MLFilter(
+        onnx_path=str(tmp_path / "nonexistent.onnx"),
+        lgbm_path=str(tmp_path / "nonexistent.pkl"),
+    )
     assert not f.is_model_loaded()
 
 
 def test_no_model_should_enter_returns_true(tmp_path):
     """모델 없으면 항상 진입 허용 (폴백)"""
-    f = MLFilter(model_path=str(tmp_path / "nonexistent.pkl"))
+    f = MLFilter(
+        onnx_path=str(tmp_path / "nonexistent.onnx"),
+        lgbm_path=str(tmp_path / "nonexistent.pkl"),
+    )
     features = make_features()
     assert f.should_enter(features) is True
 
@@ -28,7 +34,7 @@ def test_should_enter_above_threshold():
     f = MLFilter(threshold=0.60)
     mock_model = MagicMock()
     mock_model.predict_proba.return_value = np.array([[0.35, 0.65]])
-    f._model = mock_model
+    f._lgbm_model = mock_model
     features = make_features()
     assert f.should_enter(features) is True
 
@@ -38,7 +44,7 @@ def test_should_enter_below_threshold():
     f = MLFilter(threshold=0.60)
     mock_model = MagicMock()
     mock_model.predict_proba.return_value = np.array([[0.55, 0.45]])
-    f._model = mock_model
+    f._lgbm_model = mock_model
     features = make_features()
     assert f.should_enter(features) is False
 
@@ -48,8 +54,10 @@ def test_reload_model(tmp_path):
     import joblib
 
     # 모델 파일이 없는 상태에서 시작
-    model_path = tmp_path / "lgbm_filter.pkl"
-    f = MLFilter(model_path=str(model_path))
+    f = MLFilter(
+        onnx_path=str(tmp_path / "nonexistent.onnx"),
+        lgbm_path=str(tmp_path / "lgbm_filter.pkl"),
+    )
     assert not f.is_model_loaded()
 
     # _model을 직접 주입해서 is_model_loaded가 True인지 확인