Risk disclosures in annual reports and quarterly filings contain early signals about operational stress, litigation exposure, concentration risk, cyber incidents, and liquidity concerns. The difficulty for analysts is not access to documents but the ability to compare long disclosures quickly and consistently. This paper studies an applied natural language processing workflow for extracting and ranking risk signals from public financial filings. The workflow combines section segmentation, sentence-level risk scoring, and disclosure- shift analysis to track changes in risk language across reporting periods. On a labeled sample of 4,200 filing sentences, the approach achieves 86.9% precision and 83.5% recall in identifying material risk statements. A disclosure-shift index built on the extracted sentences highlights periods where banks expand discussion of fraud losses, cybersecurity, deposit pressure, or vendor dependence. The paper is framed as analyst support rather than automated investment advice.
This paper presented an applied NLP workflow for extracting risk signals from public financial filings. Using online sources and modest modeling choices, the approach improved sentence- level identification of material disclosures and supported period-over-period comparison of risk discussion. For analyst teams reviewing 2024 disclosures in early 2025, this type of filing analysis offers a narrow and defensible use of AI in finance: measurable, reviewable, and operationally useful.
References
Araci, D. (2019). FinBERT: Financial sentiment analysis with pre-trained language models.
arXiv preprint arXiv:1908.10063.
Hassan, T. A., Hollander, S., van Lent, L., and Tahoun, A. (2019). Firm-level political risk: Measurement and effects. Quarterly Journal of Economics, 134(4), 2135–2202.
Li, F. (2010). The information content of forward-looking statements in corporate filings.
Journal of Accounting Research, 48(5), 1049–1102.
Loughran, T. and McDonald, B. (2011). When is a liability not a liability? Textual analysis, dictionaries, and 10-Ks. Journal of Finance, 66(1), 35–65.
U.S. Securities and Exchange Commission. (2024). EDGAR company filings. https://www. sec.gov/edgar/search-and-access.
Beltagy, I., Peters, M. E., and Cohan, A. (2020). Longformer: The long-document transformer.
arXiv preprint arXiv:2004.05150.
How to Cite This Paper
Viswatej Seela (2025). Extracting Risk Signals from Financial Filings with Applied Natural Language Processing. International Journal of Computer Techniques, 12(1). ISSN: 2394-2231.
