---
title: Malliavin-Informed Policy Gradients for Continuous-Time Portfolio Optimization Neural Variance Reduction under Black-Scholes, Heston, and Rough Bergomi Dynamics
url: https://www.ml-quant.com/papers/ssrn/7598820/
site: ML-Quant (https://www.ml-quant.com)
updated: 2026-10-16
license: Summaries CC BY 4.0; links go to the original sources
index: https://www.ml-quant.com/llms.txt
identifier: SSRN 7598820
source_url: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7598820
featured: 2026-10-16
citations: unknown
topic: Portfolio & Allocation
---


# Malliavin-Informed Policy Gradients for Continuous-Time Portfolio Optimization Neural Variance Reduction under Black-Scholes, Heston, and Rough Bergomi Dynamics

The paper develops a Malliavin-informed policy-gradient estimator using stochastic control variates to reduce variance in continuous-time reinforcement learning for dynamic portfolio selection.

- Source: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7598820
- Identifier: SSRN 7598820
- Released: 2026-10-14
- First featured: Quant Letter No. 135 (2026-10-16): https://www.ml-quant.com/issues/2026-10-16/
- Citations (Semantic Scholar): not tracked
- Published in: not yet
- Topic: Portfolio & Allocation
- Authors: Abdou Kaba

## Related

- [Diffusion-Augmented Reinforcement Learning for Robust Portfolio Optimization under Stress Scenarios](https://www.ml-quant.com/papers/arxiv/2510.07099/): The research introduces a framework called DARL that combines DDPMs with DRL for portfolio management, improving its ability to withstand crises.
- [Adaptive and Regime-Aware RL for Portfolio Optimization](https://www.ml-quant.com/papers/arxiv/2509.14385/): The study presents a new reinforcement learning framework for portfolio optimization, which performs well under financial stress and supports dynamic asset allocation.
- [Two Stochastic Control Methods for Mean-Variance Portfolio Selection of Jump Diffusions and Their Relationship](https://www.ml-quant.com/papers/arxiv/2508.01138/): The paper examines the maximum principle and dynamic programming principle in mean-variance portfolio selection of jump diffusions, studying their interrelation.
- [Discrete-Time Mean-Variance Strategy Based on Reinforcement Learning](https://www.ml-quant.com/papers/arxiv/2312.15385/): The article discusses a new reinforcement learning-based model for analyzing real-world data, which is more applicable than the continuous-time model.
- [Evaluation of Deep Reinforcement Learning Algorithms for Portfolio Optimisation](https://www.ml-quant.com/papers/arxiv/2307.07694/): The study finds that PPO and A2C deep reinforcement learning algorithms are more effective for portfolio optimization due to their noise handling and policy derivation capabilities, despite their high sample complexity.
- [Combining Reinforcement Learning and Barrier Functions for Adaptive Risk Management in Portfolio Optimization](https://www.ml-quant.com/papers/arxiv/2306.07013/): Reinforcement learning and barrier functions used in portfolio management framework