Deep learning improves option trading by controlling risk exposure
Taming the Greeks: Option Portfolios with Inductive Biases
Machine Learning
Summary
Trading options involves managing risks that come from how prices move. The paper introduces a deep learning method that not only aims to make better returns but also carefully controls specific types of risk in the portfolio. Unlike other approaches that rely on simulating future markets, this method learns from past real data and balances making money with keeping certain risk levels low. The authors tested their method on real Nasdaq 100 options and found it led to better risk-adjusted profits and less unwanted exposure to market moves.
What this means in practice
- •For quantitative traders: Develop option trading models that improve returns while controlling directional risks using historical data.$Commercial implications: Enables creation of advanced automated options trading systems with built-in risk controls attractive to hedge funds and trading firms.
- •For financial risk managers: Assess and regulate option portfolios’ sensitivity to market movements using neural network-based risk controls.
Tested on one dataset.
Authors
Wee Ling Tan, Stephen Roberts, Stefan Zohren
Abstract
We present an end-to-end deep learning framework for systematic options trading that directly embeds hedging behavior through explicit control of portfolio-level risk exposures. While neural networks trained to optimize risk-adjusted performance have been shown to outperform traditional rules-based strategies, such approaches remain agnostic to the sensitivities of the resulting portfolios with respect to specific underlying risk factors. We propose a general training objective that combines a performance-driven loss with a differentiable risk-sensitivity penalty, enforcing neutrality to selected risk dimensions. Unlike reinforcement learning methods that approximate optimal hedging policies via simulated market dynamics, our framework operates entirely on historical data and jointly optimizes risk-adjusted returns and targeted risk constraints in a single learning problem. We instantiate the framework on static delta-neutral straddle portfolios with the penalty directed at first-order directional exposure, and evaluate two penalty variants -- an exposure-normalized penalty and a Greek-ratio drift penalty. Empirical results on Nasdaq 100 equity options demonstrate that appropriately calibrated regularization simultaneously improves out-of-sample risk-adjusted performance relative to an unregularized baseline while reducing realized directional exposure.