Generative Modeling for Trading

This unit advances students from foundational concepts to hands-on implementation of generative and hybrid models for investment and asset management. Students will translate legacy MATLAB algorithms to Python, engineer and feature-engineer financial and sentiment data, build and stabilize advanced generative architectures (VAE, flows, GAN/WGAN), and quantify model readiness for trading strategies such as options term-structure arbitrage.

Learning Objectives

• Analyze MATLAB-based trading algorithms and translate core components into functionally equivalent, tested Python implementations with appropriate documentation
• Apply data engineering and feature engineering methods to ingest, clean, transform, and label time-series, options, and sentiment datasets for use in generative and hybrid modeling pipelines
• Demonstrate design, training, and improvement of generative and hybrid models (VAE, normalizing/coupling flows, GAN/WGAN and discriminative hybrids), including model fitting, stabilization techniques, and model quantization for deployment
• Evaluate experiment results and model performance for trading and asset-management use cases by assessing calibration, risk-return metrics, backtest evidence, and readiness for operational deployment under applicable financial standards