fix(Config): rename sliding_window_size to initial_window_size (#235)

config/preprocess.py

@@ -131,7 +131,7 @@ def __preprocess_transformations_config(config_dict: dict) -> dict:
         get_scaler(config_dict["scaler"]),
         get_pca(
             config_dict["dimensionality_reduction_ratio"],
-            config_dict["sliding_window_size"],
+            config_dict["initial_window_size"],
         ),
         get_rfe(config_dict["n_features_to_select"]),
     ]

config/presets.py

@@ -16,7 +16,7 @@ def get_default_config() -> RawConfig:
     return RawConfig(
         dimensionality_reduction_ratio=0.5,
         n_features_to_select=50,
-        sliding_window_size=3800,
+        initial_window_size=3800,
         retrain_every=2000,
         scaler="minmax",  # 'normalize' 'minmax' 'standardize' 'robust'
         assets=["fivemin_crypto"],

config/sweep_ensemble.yaml

@@ -12,7 +12,7 @@ parameters:
       value: ['daily_etf']
     exogenous_data:
       value: ['daily_glassnode']
-    sliding_window_size: 
+    initial_window_size: 
       value: 380
       distribution: categorical
     n_features_to_select:

config/types.py

@@ -13,7 +13,7 @@ from transformations.base import Transformation
 class RawConfig(BaseModel):
     dimensionality_reduction_ratio: float
     n_features_to_select: int
-    sliding_window_size: int
+    initial_window_size: int
     retrain_every: int
     scaler: Literal["normalize", "minmax", "standardize", "robust"]
 
@@ -39,7 +39,7 @@ class RawConfig(BaseModel):
 
 @dataclass
 class Config:
-    sliding_window_size: int
+    initial_window_size: int
     retrain_every: int
 
     assets: DataCollection

data_loader/process.py

@@ -18,4 +18,4 @@ def check_data(X: XDataFrame, config: Config) -> bool:
 def has_enough_samples_to_train(X: XDataFrame, config: Config) -> bool:
     first_valid_index = get_first_valid_return_index(X.iloc[:, 0])
     samples_to_train = len(X) - first_valid_index
-    return samples_to_train > (config.sliding_window_size * 2) + 100
+    return samples_to_train > (config.initial_window_size * 2) + 100

training/bet_sizing.py

@@ -50,7 +50,7 @@ def bet_sizing_with_meta_model(
             X=meta_X,
             y=meta_y,
             forward_returns=forward_returns,
-            window_size=config.sliding_window_size,
+            window_size=config.initial_window_size,
             retrain_every=config.retrain_every,
             from_index=from_index,
             transformations=transformations,
@@ -62,7 +62,7 @@ def bet_sizing_with_meta_model(
         y=meta_y,
         forward_returns=forward_returns,
         model=model,
-        sliding_window_size=config.sliding_window_size,
+        initial_window_size=config.initial_window_size,
         retrain_every=config.retrain_every,
         from_index=from_index,
         level="meta",

training/directional_training.py

@@ -30,7 +30,7 @@ def train_directional_model(
             X=X,
             y=y,
             forward_returns=forward_returns,
-            window_size=config.sliding_window_size,
+            window_size=config.initial_window_size,
             retrain_every=config.retrain_every,
             from_index=from_index,
             transformations=transformations,
@@ -44,7 +44,7 @@ def train_directional_model(
         y=y,
         forward_returns=forward_returns,
         model=model,
-        sliding_window_size=config.sliding_window_size,
+        initial_window_size=config.initial_window_size,
         retrain_every=config.retrain_every,
         from_index=from_index,
         level="primary",

training/train_model.py

@@ -19,7 +19,7 @@ def train_model(
     y: pd.Series,
     forward_returns: pd.Series,
     model: Model,
-    sliding_window_size: int,
+    initial_window_size: int,
     retrain_every: int,
     from_index: Optional[pd.Timestamp],
     level: str,
@@ -42,7 +42,7 @@ def train_model(
             y=y,
             forward_returns=forward_returns,
             expanding_window=True,
-            window_size=sliding_window_size,
+            window_size=initial_window_size,
             retrain_every=retrain_every,
             from_index=from_index,
             transformations_over_time=transformations_over_time,
@@ -59,7 +59,7 @@ def train_model(
         transformations_over_time=transformations_over_time,
         X=X,
         expanding_window=True,
-        window_size=sliding_window_size,
+        window_size=initial_window_size,
         retrain_every=retrain_every,
         from_index=from_index,
     )

transformations/pca.py

@@ -10,15 +10,15 @@ class PCATransformation(Transformation):
 
     pca: PCA
 
-    def __init__(self, ratio_components_to_keep: float, sliding_window_size: int):
+    def __init__(self, ratio_components_to_keep: float, initial_window_size: int):
         self.ratio_components_to_keep = ratio_components_to_keep
-        self.sliding_window_size = sliding_window_size
+        self.initial_window_size = initial_window_size
 
     def fit(self, X: pd.DataFrame, y: Optional[pd.Series] = None) -> None:
         self.pca = PCA(
             n_components=min(
                 int(len(X.columns) * self.ratio_components_to_keep),
-                self.sliding_window_size,
+                self.initial_window_size,
             )
         )
         self.pca.fit(X, y)

transformations/retrieve.py

@@ -19,13 +19,13 @@ def get_rfe(n_feature_to_select: int) -> Optional[RFETransformation]:
 
 
 def get_pca(
-    ratio_components_to_keep: float, sliding_window_size: int
+    ratio_components_to_keep: float, initial_window_size: int
 ) -> Optional[PCATransformation]:
 
     if ratio_components_to_keep > 0:
         return PCATransformation(
             ratio_components_to_keep=ratio_components_to_keep,
-            sliding_window_size=sliding_window_size,
+            initial_window_size=initial_window_size,
         )
     else:
         return None

utils/evaluate.py

@@ -44,7 +44,7 @@ def evaluate_predictions(
     labels: list[int],
     transaction_costs: float,
 ) -> Stats:
-    # ignore the predictions until we see a non-zero returns (and definitely skip the first sliding_window_size)
+    # ignore the predictions until we see a non-zero returns (and definitely skip the first initial_window_size)
     evaluate_from = max(
         get_first_valid_return_index(forward_returns),
         get_first_valid_return_index(y_pred),