<ul><li>Researchers propose a diffusion factor model that integrates latent factor structure into generative diffusion processes for financial scenario simulation.</li><li>By exploiting the low-dimensional factor structure inherent in asset returns, the model decomposes the score function using time-varying orthogonal projections.</li><li>The proposed model provides nonasymptotic error bounds for both score estimation and generated distribution, surpassing the dimension-dependent limits in classical nonparametric statistics literature.</li><li>Numerical studies confirm superior performance in latent subspace recovery and empirical analysis demonstrates the economic significance of the framework.</li></ul>

Diffusion Factor Models: Generating High-Dimensional Returns with Factor Structure

Discover more