Backtesting

Backtesting is the process of testing a trading strategy on historical data to see how it would have performed in the past. You can easily run backtests with the framework for various timeframes and data sources.

Running a backtest

When you have created a trading strategy you can easily run a backtest in the following ways:

import logging.config
from datetime import datetime, timedelta
from investing_algorithm_framework import (
    CCXTOHLCVMarketDataSource, CCXTTickerMarketDataSource, PortfolioConfiguration, create_app, pretty_print_backtest, BacktestDateRange, TimeUnit, TradingStrategy, OrderSide, DEFAULT_LOGGING_CONFIG, Context
)

logging.config.dictConfig(DEFAULT_LOGGING_CONFIG)

bitvavo_btc_eur_ohlcv_2h = CCXTOHLCVMarketDataSource(
    identifier="BTC/EUR-ohlcv-2h",
    market="BINANCE",
    symbol="BTC/EUR",
    time_frame="2h",
    window_size=200
)

class ExampleStrategy(TradingStrategy):
    time_unit = TimeUnit.HOUR
    interval = 2
    market_data_sources = [bitvavo_btc_eur_ohlcv_2h, bitvavo_btc_eur_ticker]

    def apply_strategy(self, context: Context, market_data):
        print("Running strategy")

app = create_app(name="Example strategy")
app.add_strategy(CrossOverStrategy)

# Add a portfolio configuration of 400 euro initial balance
app.add_portfolio_configuration(
    PortfolioConfiguration(
        market="BINANCE", trading_symbol="EUR", initial_balance=400,
    )
)

if __name__ == "__main__":
    end_date = datetime(2023, 12, 2)
    start_date = end_date - timedelta(days=100)
    date_range = BacktestDateRange(
        start_date=start_date,
        end_date=end_date
    )
    backtest_report = app.run_backtest(
        backtest_date_range=date_range, save_in_memory_strategies=True
    )

Showing the backtest results

You can show the backtest results by using the pretty_print_backtest function. This will show you the backtest report in a nice format.

report = app.run_backtest(
    backtest_date_range=date_range, save_in_memory_strategies=True
)
pretty_print_backtest(report)

Loading old backtest reports

You can load old backtest reports by using the load_backtest_report function. This will load the backtest report from the file system.

from investing_algorithm_framework import load_backtest_reports

reports = load_backtest_reports(<path_to_backtest_reports_directory>)

Evaluating backtest reports

You can evaluate backtest reports by using the BacktestReportsEvaluation class. This will evaluate the backtest

from investing_algorithm_framework import BacktestReportsEvaluation, \
    pretty_print_backtest_reports_evaluation, load_backtest_reports

reports = load_backtest_reports(<path_to_backtest_reports_directory>)
evaluation = BacktestReportsEvaluation(reports)
pretty_print_backtest_reports_evaluation(evaluation)

Saving the strategy

During the backtest, the framework will not automatically save the strategy. You can save the strategy by using the save_strategy parameter of the run_backtest function. By default, it will save all the strategies that are in the strategies directory. If the directory where there strategies are located has a different name, you can specify the directory name by using the strategy_directory parameter of the run_backtest function.

The save_strategy parameter by default expects that you have the following directory structure:

.
├── app.py
├── requirements.txt
├── strategies
│   ├── data_providers.py
│   └── strategy.py
├── .gitignore
├── .env.example
└── README.md

backtest_report = app.run_backtest(backtest_date_range=date_range, save_strategy=True)

Use the strategy_directory parameter to specify the directory where the strategies are located.

backtest_report = app.run_backtest(
    backtest_date_range=date_range, save_strategy=True, strategy_directory="my_strategies"
)

If you want to directly save the loaded-in strategies, you can use the save_in_memory_strategies parameter of the run_backtest function. This will save the strategies that are loaded in memory. This is useful if you're running a backtest in a jupyter notebook where the strategy is defined in a code cell.

backtest_report = app.run_backtest(
    backtest_date_range=date_range, save_strategy=True, save_in_memory_strategies=True
)