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

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
@ -18,16 +19,11 @@ UNSET_FLOAT: float = sys.float_info.max
UNSET_INT: int = sys.maxsize
# # ==========================================================================
CONFIG = CRYPTO_CONFIG
# CONFIG = EQT_CONFIG
BacktestResults = BacktestResult(config=CONFIG)
def create_trading_signals(pair: TradingPair) -> pd.DataFrame:
result_columns = [
trades_columns = [
"time",
"action",
"symbol",
@ -35,48 +31,58 @@ def create_trading_signals(pair: TradingPair) -> pd.DataFrame:
"disequilibrium",
"scaled_disequilibrium",
"pair",
]
]
testing_pair_df = pair.testing_df_
next_values = pair.vecm_fit_.predict(steps=len(testing_pair_df))
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 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()

View File

@ -188,15 +188,9 @@ class BacktestResult:
f" {pos['open_px_b']:<8.2f}"
f" {pos['current_px_b']:<10.2f}"
f" {pos['current_value_b']:<12.2f}"
f" {'':<15}"
)
# Print pair totals with disequilibrium info
disequilibrium_status = (
"CLOSE"
if pos["current_abs_term"] < pos["closing_threshold"]
else f"{pos['disequilibrium_ratio']:.2f}x"
)
print(
f"{'':<15}"
f" {'PAIR TOTAL':<10}"
@ -205,7 +199,6 @@ class BacktestResult:
f" {'':<8}"
f" {'':<10}"
f" {pos['total_current_value']:<12.2f}"
f" {disequilibrium_status:<15}"
)
# Print disequilibrium details
@ -220,16 +213,6 @@ class BacktestResult:
f" Scaled: {pos['current_scaled_disequilibrium']:<6.4f}"
)
print(
f"{'':<15}"
f" {'THRESHOLD':<10}"
f" {'':<4}"
f" {'':<10}"
f" {'':<8}"
f" {'':<10}"
f" Close: {pos['closing_threshold']:<6.4f}"
f" Ratio: {pos['disequilibrium_ratio']:<6.2f}"
)
print("-" * 100)
total_value += pos["total_current_value"]
@ -243,7 +226,7 @@ class BacktestResult:
def handle_outstanding_position(self, pair, pair_result_df, last_row_index,
open_side_a, open_side_b, open_px_a, open_px_b,
open_tstamp, initial_abs_term, colname_a, colname_b):
open_tstamp):
"""
Handle calculation and tracking of outstanding positions when no close signal is found.
@ -254,11 +237,10 @@ class BacktestResult:
open_side_a, open_side_b: Trading sides for symbols A and B
open_px_a, open_px_b: Opening prices for symbols A and B
open_tstamp: Opening timestamp
initial_abs_term: Initial absolute disequilibrium term
colname_a, colname_b: Column names for the price data
"""
last_row = pair_result_df.loc[last_row_index]
last_tstamp = last_row["tstamp"]
colname_a, colname_b = pair.colnames()
last_px_a = last_row[colname_a]
last_px_b = last_row[colname_b]
@ -296,12 +278,9 @@ class BacktestResult:
"total_current_value": total_current_value,
"open_time": open_tstamp,
"last_time": last_tstamp,
"initial_abs_term": initial_abs_term,
"current_abs_term": current_scaled_disequilibrium,
"current_disequilibrium": current_disequilibrium,
"current_scaled_disequilibrium": current_scaled_disequilibrium,
"closing_threshold": initial_abs_term / self.config["dis-equilibrium_close_trshld"],
"disequilibrium_ratio": current_scaled_disequilibrium / (initial_abs_term / self.config["dis-equilibrium_close_trshld"]),
}
)

View File

@ -91,32 +91,6 @@ def load_market_data(datafile: str, config: Dict) -> pd.DataFrame:
return df
def transform_dataframe(df: pd.DataFrame, price_column: str):
# Select only the columns we need
df_selected = df[["tstamp", "symbol", price_column]]
# Start with unique timestamps
result_df: pd.DataFrame = pd.DataFrame(df_selected["tstamp"]).drop_duplicates().reset_index(drop=True)
# For each unique symbol, add a corresponding close price column
for symbol in df_selected["symbol"].unique():
# Filter rows for this symbol
df_symbol = df_selected[df_selected["symbol"] == symbol].reset_index(drop=True)
# Create column name like "close-COIN"
new_price_column = f"{price_column}_{symbol}"
# Create temporary dataframe with timestamp and price
temp_df = pd.DataFrame({
"tstamp": df_symbol["tstamp"],
new_price_column: df_symbol[price_column]
})
# Join with our result dataframe
result_df = pd.merge(result_df, temp_df, on="tstamp", how="left")
result_df = result_df.reset_index(drop=True) # do not dropna() since irrelevant symbol would affect dataset
return result_df
# if __name__ == "__main__":

View File

@ -4,6 +4,7 @@ import pandas as pd
from statsmodels.tsa.vector_ar.vecm import VECM
class TradingPair:
market_data_: pd.DataFrame
symbol_a_: str
symbol_b_: str
price_column_: str
@ -11,29 +12,59 @@ class TradingPair:
training_mu_: Optional[float]
training_std_: Optional[float]
original_df_: Optional[pd.DataFrame]
training_df_: Optional[pd.DataFrame]
testing_df_: Optional[pd.DataFrame]
vecm_fit_: Optional[VECM]
def __init__(self, symbol_a: str, symbol_b: str, price_column: str):
def __init__(self, market_data: pd.DataFrame, symbol_a: str, symbol_b: str, price_column: str):
self.symbol_a_ = symbol_a
self.symbol_b_ = symbol_b
self.price_column_ = price_column
self.market_data_ = self._transform_dataframe(market_data)[["tstamp"] + self.colnames()]
self.training_mu_ = None
self.training_std_ = None
self.original_df_ = None
self.training_df_ = None
self.testing_df_ = None
self.vecm_fit_ = None
def get_datasets(self, market_data: pd.DataFrame, training_minutes: int) -> None:
self.original_df_ = market_data[["tstamp"] + self.colnames()]
self.training_df_ = market_data.iloc[:training_minutes - 1, :].copy()
def _transform_dataframe(self, df: pd.DataFrame):
# Select only the columns we need
df_selected = df[["tstamp", "symbol", self.price_column_]]
# Start with unique timestamps
result_df: pd.DataFrame = pd.DataFrame(df_selected["tstamp"]).drop_duplicates().reset_index(drop=True)
# For each unique symbol, add a corresponding close price column
for symbol in df_selected["symbol"].unique():
# Filter rows for this symbol
df_symbol = df_selected[df_selected["symbol"] == symbol].reset_index(drop=True)
# Create column name like "close-COIN"
new_price_column = f"{self.price_column_}_{symbol}"
# Create temporary dataframe with timestamp and price
temp_df = pd.DataFrame({
"tstamp": df_symbol["tstamp"],
new_price_column: df_symbol[self.price_column_]
})
# Join with our result dataframe
result_df = pd.merge(result_df, temp_df, on="tstamp", how="left")
result_df = result_df.reset_index(drop=True) # do not dropna() since irrelevant symbol would affect dataset
return result_df
def get_datasets(self, training_minutes: int, training_start_index: int = 0, testing_size: Optional[int] = None) -> None:
self.training_df_ = self.market_data_.iloc[training_start_index:training_minutes - 1, :].copy()
self.training_df_ = self.training_df_.dropna().reset_index(drop=True)
self.testing_df_ = market_data.iloc[training_minutes:, :].copy()
testing_start_index = training_start_index + training_minutes
if testing_size is None:
self.testing_df_ = self.market_data_.iloc[testing_start_index:, :].copy()
else:
self.testing_df_ = self.market_data_.iloc[testing_start_index:testing_start_index + testing_size, :].copy()
self.testing_df_ = self.testing_df_.dropna().reset_index(drop=True)
def colnames(self) -> List[str]:
@ -70,7 +101,7 @@ class TradingPair:
return False
pass
print(f"*****\n**************** {self} IS COINTEGRATED ****************\n*****")
print('*' * 80 + '\n' + f"**************** {self} IS COINTEGRATED ****************\n" + '*' * 80)
self.fit_VECM()
diseq_series = self.training_df_[self.colnames()] @ self.vecm_fit_.beta
self.training_mu_ = diseq_series.mean().iloc[0]
@ -84,6 +115,31 @@ class TradingPair:
return True
def predict(self) -> None:
predicted_prices = self.vecm_fit_.predict(steps=len(self.testing_df_))
# Convert prediction to a DataFrame for readability
# predicted_df =
self.predicted_df_ = pd.merge(
self.testing_df_.reset_index(drop=True),
pd.DataFrame(predicted_prices, columns=self.colnames()),
left_index=True,
right_index=True,
suffixes=("", "_pred"),
).dropna()
self.predicted_df_["disequilibrium"] = self.predicted_df_[self.colnames()] @ self.vecm_fit_.beta
self.predicted_df_["scaled_disequilibrium"] = (
abs(self.predicted_df_["disequilibrium"] - self.training_mu_) / self.training_std_
)
# Reset index to ensure proper indexing
self.predicted_df_ = self.predicted_df_.reset_index()
return self.predicted_df_
def __repr__(self) ->str:
return f"{self.symbol_a_} & {self.symbol_b_}"