---
title: Dynamic data generation and dynamic portfolio selection: an application of a score-based diffusion model
url: https://www.ml-quant.com/papers/arxiv/2507.09916/
site: ML-Quant (https://www.ml-quant.com)
updated: 2026-09-26
license: Summaries CC BY 4.0; links go to the original sources
index: https://www.ml-quant.com/llms.txt
identifier: arXiv:2507.09916
source_url: http://arxiv.org/abs/2507.09916v1
featured: 2025-07-17
citations: 5
topic: Portfolio & Allocation
---


# Dynamic data generation and dynamic portfolio selection: an application of a score-based diffusion model

A model-free approach using generative diffusion models and a policy gradient algorithm has been suggested for dynamic mean-variance portfolio selection, outperforming the Markowitz portfolio and S&P 500.

- Source: http://arxiv.org/abs/2507.09916v1
- Identifier: arXiv:2507.09916
- Released: 2025-07-14
- First featured: Quant Letter No. 106 (2025-07-17): https://www.ml-quant.com/issues/2025-07-17/
- Citations (Semantic Scholar): 5
- Published in: not yet
- Topic: Portfolio & Allocation

## Related

- [A monotone numerical integration method for mean-variance portfolio optimization under jump-diffusion models](https://www.ml-quant.com/papers/arxiv/2309.05977/): The research introduces an efficient numerical integration method for portfolio optimization, demonstrating its computational efficiency and accuracy, and its convergence to the unique solution of the optimization problem.
- [Improving Portfolio Construction with Synthetic Time Series Data](https://www.ml-quant.com/papers/ssrn/4809931/): A deep generative model can create high-quality, correlation-aware synthetic financial time series, useful in a GAN-assisted risk-based portfolio construction method.
- [Quantum stochastic walks for portfolio optimization: theory and implementation on financial networks](https://www.ml-quant.com/papers/arxiv/2507.03963/): A study suggests a quantum stochastic walk optimizer for financial markets, which increases the annualized Sharpe ratio by 15% and reduces turnover by 90% compared to traditional optimization.
- [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.
- [Estimating Covariance for Global Minimum Variance Portfolio: A Decision-Focused Learning Approach](https://www.ml-quant.com/papers/arxiv/2508.10776/): The paper introduces a decision-focused learning approach for portfolio optimization, which improves decision performance compared to prediction-focused methods.
- [THEME: Enhancing Thematic Investing with Semantic Stock Representations and Temporal Dynamics](https://www.ml-quant.com/papers/arxiv/2508.16936/): Enhancing Thematic Investing: The study presents THEME, a hierarchical contrastive learning framework for thematic investing, which surpasses baselines in multiple retrieval metrics and enhances portfolio construction performance.