Skip to content

Welcome to Quantreo

A faster way to build quantitative features, targets, and alternative bars in Python. Powered by Numba.

PyPI Python versions Downloads License Last Commit


Quantreo provides a high-performance research framework for building features, targets, and alternative bars in Python. Its goal is to help quantitative researchers transform raw market data into machine-learning-ready datasets quickly, efficiently, and without data leakage.

Each module is built around the principles of clarity, reproducibility, and performance, using Numba for just-in-time optimization. Whether your goal is alpha research, backtesting, or production-ready model development, Quantreo offers a consistent and modular workflow.

Main Packages

Quantreo is built for quantitative research, Numba-optimized, leakage-safe, and fully compatible with pandas and scikit-learn, ensuring fast and reproducible data transformations across all modules.

Package Purpose Example Capabilities
Data Aggregation Transform raw tick data into structured OHLCV bars Time bars, tick bars, volume bars, run bars
Features Engineering Extract predictive information from price and volume Volatility, entropy, trend, statistical structure
Target Engineering Build machine-learning targets without data leakage Triple-barrier, meta-labeling, event-based labeling

Each module can be used independently or combined into a complete research pipeline, from raw data to model-ready datasets.

Installation

Quantreo is available on PyPI and can be installed with pip:

pip install quantreo

To verify your installation:

import quantreo
print(quantreo.__version__)

Quick Start

Compute a ready-to-use volatility feature in just two lines.

import quantreo.features_engineering as fe

df["parkinson_vol"] = fe.volatility.parkinson_volatility(df=df, high_col="high",
                                                         low_col="low", window_size=30)
Then, you can easily visualize your computed feature to check its behavior over time:

Parkinson Volatility Feature example

Real-Life Applications

Quantreo is not just a feature library, it’s designed for real research workflows. Explore practical examples demonstrating how to apply Quantreo in end-to-end trading research pipelines.

Each notebook showcases real-world use cases of Quantreo, from feature design to model-ready data generation.

Example Description
Meta-Labelling Explained Learn how to apply meta-labelling to improve signal precision and reduce false positives in trading strategies.
Multi-Asset Feature Engineering in Financial ML Discover how to build and standardize multi-asset features for cross-asset modeling and synthetic dataset generation.
Dimensionality Reduction in Trading Use Kernel PCA to create synthetic volatility features and reduce feature space complexity in ML models.

Troubleshooting & Support

If you encounter any issue with installation or usage, please don’t hesitate to reach out.

You can:

All feedback and bug reports are welcome, they help improve Quantreo’s performance, reliability, and usability for the quantitative research community.