DeepFolio is an OpenAI Sora-inspired Diffusion-Transformer (DiT) framework for joint portfolio optimization and best execution, designed to maximize Sharpe ratio without explicit return forecasts. It leverages:
- Transformer to capture asset dependencies and encode market conditions.
- Diffusion Models to filter market noise and generate both robust allocation weights and optimized trading trajectories.
- End-to-End Strategy Execution to reduce information loss between strategy design and execution implementation, ensuring optimal real-world performance.
β
Unified Portfolio & Execution Optimization β Bridges the gap between portfolio construction and trade execution.
β
Diffusion-Based Portfolio Generation β Generates adaptive, robust asset allocations without relying on explicit return forecasts.
β
Market-Aware Execution Path Modeling β Uses Diffusion Models to optimize execution trajectories, reducing slippage and market impact.
β
Scenario-Based Adaptation β Dynamically adjusts strategies for high/low volatility regimes, liquidity shifts, and market anomalies.
β
Transaction Cost-Aware Optimization β Integrates TCA (Transaction Cost Analysis) into optimization, minimizing execution costs.
DeepFolio consists of two core modules:
- Transformer Encoder extracts asset relationships, learning market structure.
- Diffusion Model generates optimal portfolio weights, ensuring robustness under different conditions.
- Transformer encodes market microstructure (LOB, liquidity, volatility).
- Diffusion Model optimizes execution paths to minimize market impact and transaction costs.
π Pipeline Overview:
For detailed documentation, please visit our documentation site.
We welcome contributions! Please see our contributing guidelines for more details.
This project is licensed under the BSD-2-Clause License- see the LICENSE file for details.
[1] Damian Kisiel, Denise Gorse (2022). Portfolio Transformer for Attention-Based Asset Allocation arXiv:2206.03246 [q-fin.PM]
- This package leverages the power of TensorFlow for efficient portfolio optimization.
- Thanks to the financial machine learning community for inspiring many of the implemented methods.