Backtest Optimizer

Optimization of trading strategy hyperparameters with combinatorial cross validation and stress tesing

Table of Contents

• Installation
• Usage
• Contributing
• License
• Contact

Installation

To get a local copy up and running, follow these simple steps:

Prerequisites

Ensure you have the following software installed on your system:

• Python: You can download it from python.org.
• Git: You can download it from git-scm.com.

Install the Repository

pip install git+https://github.com/alex33d/backtest_optimizer.git

Usage

To start using the Optimizer, you need to import the ParameterOptimizer class from backtest_optimizer/main.py. The main functions are:

1. Initialize the Optimizer:

   from backtest_optimizer.main import ParameterOptimizer
   
   optimizer = ParameterOptimizer(calc_pl)

   • calc_pl: A function for calculating strategy profit and loss. It must accept a dictionary with data in the form {ticker: DataFrame}, and returns pandas Series with returns for each period.

2. Train/Test Split:

   optimizer.split_data(data_dict, train_end)

   • data_dict: Dictionary containing data in the form {ticker: DataFrame}.
   • train_end: The train end date in the form of a pandas DateTime.

3. Create Combinatorial Cross Validation Groups:

   optimizer.create_combcv_dict(n_splits, n_test_splits)

   • n_splits: Total number of groups.
   • n_test_splits: Number of test groups.

4. Run the Optimization Algorithm:

   optimizer.optimize(params=params_dict, n_jobs=3, n_runs=10, best_trials_pct=0.25)

   • params: The parameter dictionary with hyperparameter space for Optuna optimization.
   • n_jobs: Number of parallel workers.
   • n_runs: Number of Optuna trials.
   • best_trials_pct: Percentage of trials that go to the validation set.

5. Reconstruct Equity Curves:

   optimizer.reconstruct_equity_curves()

   • Reconstructs the equity curves based on validation sets.

6. Run Stress Tests:

   optimizer.run_stress_tests()

   • Runs backtest stress tests.

7. Select Best Parameters:

   best_params = optimizer.cluster_and_aggregate()

   • Selects the best parameters via clustering.

8. Calculate Out-of-Sample Sharpe Ratio:

   optimizer.calc_oos_sharpe(best_params)

   • Tests the best parameters on the test set.

Example Usage

Here’s a complete example of how you might use the Optimizer class:

from backtest_optimizer.main import ParameterOptimizer

# Define your profit and loss calculation function
def calc_pl(data):
    # Calculate P&L logic here
    pass

# Initialize the optimizer
optimizer = ParameterOptimizer(calc_pl)

# Split data into train and test sets
data_dict = {'AAPL': data_frame_aapl, 'GOOG': data_frame_goog}
train_end = pd.to_datetime('2023-01-01')
optimizer.split_data(data_dict, train_end)

# Create combinatorial cross-validation groups
optimizer.create_combcv_dict(n_splits=5, n_test_splits=1)

# Define parameter dictionary for optimization
params_dict = {
    'param1': [0, 1, 2],
    'param2': [0, 10, 20]
}

# Run optimization
optimizer.optimize(params=params_dict, n_jobs=3, n_runs=10, best_trials_pct=0.25)

# Reconstruct equity curves
optimizer.reconstruct_equity_curves()

# Run stress tests
optimizer.run_stress_tests()

# Select best parameters
best_params = optimizer.cluster_and_aggregate()

# Calculate out-of-sample Sharpe ratio
oos_sharpe = optimizer.calc_oos_sharpe(best_params)

Contributing

I welcome contributions to the project! Here’s how you can help:

Reporting Issues

If you find a bug or have a feature request, please create an issue on GitHub.

1. Go to the Issues page.
2. Click on the "New issue" button.
3. Provide a clear and descriptive title.
4. Describe the issue or feature request in detail, including steps to reproduce the issue if applicable.

Submitting Pull Requests

If you want to contribute code, follow these steps to submit a pull request:

1. Fork the Repository

   Click the "Fork" button at the top right of the repository page to create a copy of the repository on your GitHub account.

2. Clone Your Fork

   Clone your forked repository to your local machine:

   git clone https://github.com/alex33d/backtest_optimizer.git
   cd backtest_optimizer

3. Create a Branch

   Create a new branch for your feature or bug fix:

   git checkout -b my-feature-branch

4. Make Your Changes

   Make your changes to the codebase. Ensure your code follows the project's coding standards.

5. Commit Your Changes

   Commit your changes with a descriptive commit message:

   git add .
   git commit -m "Description of your changes"

6. Push to Your Fork

   Push your changes to your forked repository:

   git push origin my-feature-branch

7. Create a Pull Request

   Go to the original repository on GitHub and click the "New pull request" button.

   • Select your branch from the "compare" dropdown.
   • Provide a clear and descriptive title for your pull request.
   • Describe your changes in detail in the description field.
   • Submit the pull request.

Coding Guidelines

To maintain consistency in the codebase, please adhere to the following guidelines:

• Follow the PEP 8 style guide for Python code.
• Write descriptive commit messages.
• Include docstrings for all functions and classes.
• Write unit tests for new features and bug fixes.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Contact

For any questions, suggestions, or feedback, feel free to contact me:

• LinkedIn: https://www.linkedin.com/in/alexander-demachev-b067759b/