DisclosureBeta: A Measurement-Channel Theory for Regime-Conditioned Betas from LLM-Read Risk Disclosures
The problem is the beta a desk needs when a firm's price history is too short to trust: an S-1 filer, a recent listing, or a name just past a regime break. The state of the art collapses to a comparable-firm peer beta with no error budget, and the recent text-based competitor Breitung (2025) reports strong empirical IPO accuracy but no identification theory, no error budget, and no lower bound. We fill that gap. We model a large language model as a noisy measurement channel on a firm's latent risk characteristics and write its channel noise into the asset-pricing error budget. In a piecewise-stationary Fama-French five-factor model the loadings are a function of latent risk characteristic...
| Authors | Ping Kuen Wong |
|---|---|
| Primary category | q-fin.RM |
| Published | 5 July 2026 |
| Last updated at source | 5 July 2026 |
Silytics shows public metadata from arXiv and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.