progress

src/pt_backtest.py

@@ -8,7 +8,7 @@ import numpy as np
 # ============= statsmodels ===================
 from statsmodels.tsa.vector_ar.vecm import VECM
 
-from tools.data_loader import get_datasets, load_market_data, transform_dataframe
+from tools.data_loader import load_market_data, transform_dataframe
 from tools.trading_pair import TradingPair
 from results import BacktestResult
 
@@ -88,8 +88,8 @@ EQT_CONFIG: Dict = {
     "price_column": "close",
     "min_required_points": 30,
     "zero_threshold": 1e-10,
-    "disequilibrium_open_trshld": 5.0,
-    "disequilibrium_close_trshld": 1.0,
+    "disequilibrium_open_trshld": 2.0,
+    "disequilibrium_close_trshld": 0.5,
     "training_minutes": 120,
     # ----- Validation
     "funding_per_pair": 2000.0,
@@ -104,23 +104,7 @@ CONFIG = EQT_CONFIG
 BacktestResults = BacktestResult(config=CONFIG)
 
 
-def fit_VECM(training_pair_df, pair: TradingPair):
-    vecm_model = VECM(
-        training_pair_df[pair.colnames()].reset_index(drop=True), coint_rank=1
-    )
-    vecm_fit = vecm_model.fit()
-
-    # Check if the model converged properly
-    if not hasattr(vecm_fit, "beta") or vecm_fit.beta is None:
-        print(f"{pair}: VECM model failed to converge properly")
-
-    return vecm_fit
-
-
-
-def create_trading_signals(
-    vecm_fit, testing_pair_df, pair: TradingPair
-) -> pd.DataFrame:
+def create_trading_signals(pair: TradingPair) -> pd.DataFrame:
     result_columns = [
         "time",
         "action",
@@ -131,13 +115,14 @@ def create_trading_signals(
         "pair",
     ]
 
-    next_values = vecm_fit.predict(steps=len(testing_pair_df))
+    testing_pair_df = pair.testing_df_
+    next_values = pair.vecm_fit_.predict(steps=len(testing_pair_df))
     colname_a, colname_b = pair.colnames()
 
     # Convert prediction to a DataFrame for readability
     predicted_df = pd.DataFrame(next_values, columns=[colname_a, colname_b])
 
-    beta = vecm_fit.beta
+    beta = pair.vecm_fit_.beta
 
     pair_result_df = pd.merge(
         testing_pair_df.reset_index(drop=True),
@@ -149,12 +134,9 @@ def create_trading_signals(
 
     pair_result_df["disequilibrium"] = pair_result_df[pair.colnames()] @ beta
 
-    pair_mu = pair.disequilibrium_mu_
-    pair_std = pair.disequilibrium_std_
-
     pair_result_df["scaled_disequilibrium"] = abs(
-        pair_result_df["disequilibrium"] - pair_mu
-    ) / pair_std
+        pair_result_df["disequilibrium"] - pair.training_mu_
+    ) / pair.training_std_
 
 
     # Reset index to ensure proper indexing
@@ -311,54 +293,19 @@ def create_trading_signals(
 
 
 def run_single_pair(
-    market_data: pd.DataFrame, price_column: str, pair: TradingPair
+    pair: TradingPair, market_data: pd.DataFrame, price_column: str
 ) -> Optional[pd.DataFrame]:
-    training_pair_df, testing_pair_df = get_datasets(
-        df=market_data, training_minutes=CONFIG["training_minutes"], pair=pair
+    pair.get_datasets(
+        market_data=market_data, training_minutes=CONFIG["training_minutes"]
     )
-
-    # Check if we have enough data points for a meaningful analysis
-    min_required_points = CONFIG[
-        "min_required_points"
-    ]  # Minimum number of points for a reasonable VECM model
-    if len(training_pair_df) < min_required_points:
-        print(
-            f"{pair}: Not enough data points for analysis. Found {len(training_pair_df)}, need at least {min_required_points}"
-        )
-        return None
-
-    # Check for non-finite values
-    if not np.isfinite(training_pair_df).all().all():
-        print(f"{pair}: Data contains non-finite values (NaN or inf)")
-        return None
-
-    # Fit the VECM
     try:
-        vecm_fit = fit_VECM(training_pair_df, pair=pair)
+        pair.train_pair()
     except Exception as e:
-        print(f"{pair}: VECM fitting failed: {str(e)}")
+        print(f"{pair}: Training failed: {str(e)}")
         return None
 
-    # Add safeguard against division by zero
-    if (
-        abs(vecm_fit.beta[1]) < CONFIG["zero_threshold"]
-    ):  # Small threshold to avoid division by very small numbers
-        print(f"{pair}: Skipping due to near-zero beta[1] value: {vecm_fit.beta[1]}")
-        return None
-    diseqlbrm_series = training_pair_df[pair.colnames()] @ vecm_fit.beta
-    diseqlbrm_series_mu: float = diseqlbrm_series.mean().iloc[0]
-    diseqlbrm_series_std: float = diseqlbrm_series.std().iloc[0]
-    pair.set_training_disequilibrium(diseqlbrm_series_mu, diseqlbrm_series_std)
-
-    # Normalize the disequilibrium
-    training_pair_df["scaled_disequilibrium"] = (
-        diseqlbrm_series - diseqlbrm_series_mu
-    ) / diseqlbrm_series_std
-
     try:
         pair_trades = create_trading_signals(
-            vecm_fit=vecm_fit,
-            testing_pair_df=testing_pair_df,
             pair=pair,
         )
     except Exception as e:
@@ -385,18 +332,10 @@ def run_pairs(config: Dict, market_data_df: pd.DataFrame, price_column: str) ->
 
     pairs_trades = []
     for pair in _create_pairs(config):
-        # Get the actual variable names
-        # colname_a = stock_price_columns[a_index]
-        # colname_b = stock_price_columns[b_index]
-
-        # symbol_a = colname_a[len(f"{price_column}-") :]
-        # symbol_b = colname_b[len(f"{price_column}-") :]
-        # pair = TradingPair(symbol_a, symbol_b, price_column)
-
         single_pair_trades = run_single_pair(
             market_data=market_data_df, price_column=price_column, pair=pair
         )
-        if len(single_pair_trades) > 0:
+        if single_pair_trades is not None and len(single_pair_trades) > 0:
             pairs_trades.append(single_pair_trades)
     # Check if result_list has any data before concatenating
     if len(pairs_trades) == 0:
@@ -441,10 +380,7 @@ if __name__ == "__main__":
 
             print(f"Successfully processed {filename}")
 
-            # Print total unrealized PnL for this file
-            print(
-                f"\n====== TOTAL UNREALIZED PnL for {filename}: {BacktestResults.get_total_unrealized_pnl():.2f}% ======"
-            )
+            # No longer printing unrealized PnL since we removed that functionality
 
         except Exception as e:
             print(f"Error processing {datafile}: {str(e)}")