research/stat_pairs_backtest
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

progress

Browse Source
This commit is contained in:
Oleg Sheynin
2025-05-29 15:47:56 -04:00
parent 91623db4b7
commit 50674bd3b8
4 changed files with 144 additions and 152 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/pt_backtest.py
+78 -95
View File
@@ -1,3 +1,4 @@
from abc import ABC, abstractmethod
import sys
from typing import Any, Dict, List, Optional
@@ -9,7 +10,7 @@ import numpy as np
from statsmodels.tsa.vector_ar.vecm import VECM
from backtest_configs import CRYPTO_CONFIG
from tools.data_loader import load_market_data, transform_dataframe
from tools.data_loader import load_market_data
from tools.trading_pair import TradingPair
from results import BacktestResult
@@ -18,65 +19,70 @@ UNSET_FLOAT: float = sys.float_info.max
UNSET_INT: int = sys.maxsize
# # ==========================================================================
CONFIG = CRYPTO_CONFIG
# CONFIG = EQT_CONFIG
trades_columns = [
"time",
"action",
"symbol",
"price",
"disequilibrium",
"scaled_disequilibrium",
"pair",
]
BacktestResults = BacktestResult(config=CONFIG)
class PairTradingStrategy(ABC):
@abstractmethod
def create_trading_signals(pair: TradingPair, config: Dict) -> pd.DataFrame:
...
@abstractmethod
def run_pair(pair: TradingPair) -> Optional[pd.DataFrame]:
...
def create_trading_signals(pair: TradingPair) -> pd.DataFrame:
result_columns = [
"time",
"action",
"symbol",
"price",
"disequilibrium",
"scaled_disequilibrium",
"pair",
]
testing_pair_df = pair.testing_df_
next_values = pair.vecm_fit_.predict(steps=len(testing_pair_df))
def run_pair(pair: TradingPair) -> Optional[pd.DataFrame]:
pair.get_datasets(training_minutes=CONFIG["training_minutes"])
try:
is_cointegrated = pair.train_pair()
if not is_cointegrated:
print(f"{pair} IS NOT COINTEGRATED")
return None
except Exception as e:
print(f"{pair}: Training failed: {str(e)}")
return None
try:
pair.predict()
except Exception as e:
print(f"{pair}: Prediction failed: {str(e)}")
return None
pair_trades = create_trading_signals(pair=pair, config=CONFIG)
return pair_trades
def create_trading_signals(pair: TradingPair, config: Dict) -> pd.DataFrame:
beta = pair.vecm_fit_.beta
colname_a, colname_b = pair.colnames()
# Convert prediction to a DataFrame for readability
predicted_df = pd.DataFrame(next_values, columns=[colname_a, colname_b])
predicted_df = pair.predicted_df_
beta = pair.vecm_fit_.beta
pair_result_df = pd.merge(
testing_pair_df.reset_index(drop=True),
predicted_df,
left_index=True,
right_index=True,
suffixes=("", "_pred"),
).dropna()
pair_result_df["disequilibrium"] = pair_result_df[pair.colnames()] @ beta
pair_result_df["scaled_disequilibrium"] = abs(
pair_result_df["disequilibrium"] - pair.training_mu_
) / pair.training_std_
# Reset index to ensure proper indexing
pair_result_df = pair_result_df.reset_index()
open_threshold = config["dis-equilibrium_open_trshld"]
close_threshold = config["dis-equilibrium_close_trshld"]
# Iterate through the testing dataset to find the first trading opportunity
open_row_index = None
initial_abs_term = None
open_threshold = CONFIG["dis-equilibrium_open_trshld"]
close_threshold = CONFIG["dis-equilibrium_close_trshld"]
for row_idx in range(len(pair_result_df)):
curr_disequilibrium = pair_result_df["scaled_disequilibrium"][row_idx]
for row_idx in range(len(predicted_df)):
curr_disequilibrium = predicted_df["scaled_disequilibrium"][row_idx]
# Check if current row has sufficient disequilibrium (not near-zero)
if curr_disequilibrium >= open_threshold:
open_row_index = row_idx
initial_abs_term = curr_disequilibrium
break
# If no row with sufficient disequilibrium found, skip this pair
@@ -85,7 +91,9 @@ def create_trading_signals(pair: TradingPair) -> pd.DataFrame:
return pd.DataFrame()
# Look for close signal starting from the open position
trading_signals_df = (pair_result_df["scaled_disequilibrium"][open_row_index:] < close_threshold)
trading_signals_df = (
predicted_df["scaled_disequilibrium"][open_row_index:] < close_threshold
)
# Adjust indices to account for the offset from open_row_index
close_row_index = None
@@ -94,7 +102,7 @@ def create_trading_signals(pair: TradingPair) -> pd.DataFrame:
close_row_index = idx
break
open_row = pair_result_df.loc[open_row_index]
open_row = predicted_df.loc[open_row_index]
open_tstamp = open_row["tstamp"]
open_disequilibrium = open_row["disequilibrium"]
open_scaled_disequilibrium = open_row["scaled_disequilibrium"]
@@ -102,8 +110,8 @@ def create_trading_signals(pair: TradingPair) -> pd.DataFrame:
open_px_b = open_row[f"{colname_b}"]
abs_beta = abs(beta[1])
pred_px_b = pair_result_df.loc[open_row_index][f"{colname_b}_pred"]
pred_px_a = pair_result_df.loc[open_row_index][f"{colname_a}_pred"]
pred_px_b = predicted_df.loc[open_row_index][f"{colname_b}_pred"]
pred_px_a = predicted_df.loc[open_row_index][f"{colname_a}_pred"]
if pred_px_b * abs_beta - pred_px_a > 0:
open_side_a = "BUY"
@@ -119,21 +127,18 @@ def create_trading_signals(pair: TradingPair) -> pd.DataFrame:
# If no close signal found, print position and unrealized PnL
if close_row_index is None:
last_row_index = len(pair_result_df) - 1
last_row_index = len(predicted_df) - 1
# Use the new method from BacktestResult to handle outstanding positions
BacktestResults.handle_outstanding_position(
pair=pair,
pair_result_df=pair_result_df,
pair_result_df=predicted_df,
last_row_index=last_row_index,
open_side_a=open_side_a,
open_side_b=open_side_b,
open_px_a=open_px_a,
open_px_b=open_px_b,
open_tstamp=open_tstamp,
initial_abs_term=initial_abs_term,
colname_a=colname_a,
colname_b=colname_b
)
# Return only open trades (no close trades)
@@ -159,7 +164,7 @@ def create_trading_signals(pair: TradingPair) -> pd.DataFrame:
]
else:
# Close signal found - create complete trade
close_row = pair_result_df.loc[close_row_index]
close_row = predicted_df.loc[close_row_index]
close_tstamp = close_row["tstamp"]
close_disequilibrium = close_row["disequilibrium"]
close_scaled_disequilibrium = close_row["scaled_disequilibrium"]
@@ -210,56 +215,35 @@ def create_trading_signals(pair: TradingPair) -> pd.DataFrame:
# Add tuples to data frame
return pd.DataFrame(
trd_signal_tuples,
columns=result_columns,
columns=trades_columns,
)
def run_single_pair(
pair: TradingPair, market_data: pd.DataFrame, price_column: str
) -> Optional[pd.DataFrame]:
pair.get_datasets(
market_data=market_data, training_minutes=CONFIG["training_minutes"]
)
try:
is_cointegrated = pair.train_pair()
if not is_cointegrated:
print(f"{pair} IS NOT COINTEGRATED")
return None
except Exception as e:
print(f"{pair}: Training failed: {str(e)}")
return None
try:
pair_trades = create_trading_signals(
pair=pair,
)
except Exception as e:
print(f"{pair}: Prediction failed: {str(e)}")
return None
return pair_trades
def run_pairs(config: Dict, market_data_df: pd.DataFrame, price_column: str) -> None:
def run_all_pairs(config: Dict, datafile: str, price_column: str) -> None:
def _create_pairs(config: Dict) -> List[TradingPair]:
nonlocal datafile
instruments = config["instruments"]
all_indexes = range(len(instruments))
unique_index_pairs = [(i, j) for i in all_indexes for j in all_indexes if i < j]
pairs = []
market_data_df = load_market_data(
f'{config["data_directory"]}/{datafile}', config=CONFIG
)
for a_index, b_index in unique_index_pairs:
symbol_a = instruments[a_index]
symbol_b = instruments[b_index]
pair = TradingPair(symbol_a, symbol_b, price_column)
pair = TradingPair(
market_data=market_data_df,
symbol_a=instruments[a_index],
symbol_b=instruments[b_index],
price_column=price_column,
)
pairs.append(pair)
return pairs
pairs_trades = []
for pair in _create_pairs(config):
single_pair_trades = run_single_pair(
market_data=market_data_df, price_column=price_column, pair=pair
)
single_pair_trades = run_pair(pair=pair)
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
@@ -275,7 +259,7 @@ def run_pairs(config: Dict, market_data_df: pd.DataFrame, price_column: str) ->
# BacktestResults.print_single_day_results()
if __name__ == "__main__":
def main() -> None:
# Initialize a dictionary to store all trade results
all_results: Dict[str, Dict[str, Any]] = {}
@@ -291,13 +275,9 @@ if __name__ == "__main__":
# Process data for this file
try:
market_data_df = load_market_data(
f'{CONFIG["data_directory"]}/{datafile}', config=CONFIG
run_all_pairs(
config=CONFIG, datafile=datafile, price_column=price_column
)
market_data_df = transform_dataframe(
df=market_data_df, price_column=price_column
)
run_pairs(config=CONFIG, market_data_df=market_data_df, price_column=price_column)
# Store results with file name as key
filename = datafile.split("/")[-1]
@@ -315,3 +295,6 @@ if __name__ == "__main__":
# Print grand totals
BacktestResults.print_grand_totals()
BacktestResults.print_outstanding_positions()
if __name__ == "__main__":
main()