EDGAR Risk-to-MD&A Sentiment Divergence →

An equity-research agent polls ADW-208 weekly for each holding in a fundamental portfolio; when divergence_score exceeds 1.5 standard deviations (risk_neg_density sharply above mda_neg_density, indicating management is burying bad news in the boilerplate Risk Factors while keeping MD&A upbeat), the agent flags the position for immediate review and drafts a sell-side alert citing the specific Loughran-McDonald density gap. The source_lineage pointing directly to SEC EDGAR full-text and the frozen methodology_version (Loughran-McDonald lexicon) give compliance a clean audit trail — no black-box sentiment API, just traceable word counts from the official filing. With a documented section-differential IC of 0.06–0.10, the signal carries statistically meaningful predictive weight that the agent can cite when escalating to a human analyst.

A long/short equity PM uses ADW-208 at the time of each 10-K or 10-Q filing to catch the 'management tone gap' before earnings calls amplify or dismiss it. Instead of reading hundreds of pages of filings, the PM receives a single divergence_score and the underlying density numbers, letting her zero in on the two or three names where the Risk-Factors section is markedly more negative than the MD&A — a pattern historically associated with earnings disappointment. This replaces a manual analyst process that previously flagged only the most egregious cases, turning a monthly spot-check into a continuous weekly screen across the entire coverage universe.