quant research // summarizing the paperLOADING
Rian Dolphin, Joe Dursun, Jarrett Blankenship, Katie Adams, Quinton Pike · 2026-07-09
The paper builds a two-stage large-language-model system that classifies U.S. SEC 8-K filings into a detailed taxonomy of 119 event types, anchoring each label to a verbatim quote and scoring its reliability. Applied to ~293,000 filings from 2022–2026, it produces ~601,000 grounded event tags (released publicly), with precision rising from 12% to 96% as the quality score increases. An event study on abnormal returns shows the fine-grained taxonomy distinguishes economically different events that share the same coarse SEC item code.
Why it matters: 8-K filings signal material corporate events, but the SEC's coarse item codes lump together trivial and market-moving disclosures; this finer, source-traceable labeling could help event-driven or news-based strategies isolate the disclosures that actually matter. The released dataset of grounded event tags may be a useful input for building or backtesting event-study and signal-extraction pipelines.
⚠ The paper validates labeling accuracy and event-return separation, not any trading profitability, and the LLM tagging still has variable precision requiring careful score-based filtering.
Form 8-K filings are the primary channel through which U.S. public companies disclose material events, but the SEC item codes attached to them are coarse: a single item spans routine administrative changes and chief executive departures, and many of the most market-moving disclosures fall into a catch-all item. Large language models make fine-grained labelling feasible at corpus scale, but only if the labels can be traced to the source text and shown to be reliable. We present a two-stage system that tags 8-K disclosures against a three-tier taxonomy of 119 event types. The first stage constrains output to valid taxonomy entries and anchors every tag to a verbatim quote via fuzzy n-gram validation; the second re-grades each cited quote against the category definition to produce a quality score. Applying the system to 292,984 filings from 2022 to 2026 yields 601,088 grounded event tags, which we release. Over 5,125 stratified tags, an LLM judge finds precision rises monotonically with the quality score, from 12% to 96%, while unsupported tags fall from 8% to near zero. Ablation shows the score is calibrated only when assigned in a dedicated second pass. An event study on unsigned abnormal returns confirms, without any language model, that the taxonomy separates economically distinct events sharing an item code.
Go deeper: a full research-committee breakdown of this paper, its assumptions and failure modes, and how its method would apply to a specific ticker or your watchlist. See StockTools AI →
AI summary generated from the paper’s public abstract via arXiv; it may miss nuance — read the source before relying on it. Thank you to arXiv for its open-access interoperability; StockTools is not affiliated with arXiv, and all rights remain with the authors. Educational only, not financial advice.