Wiseek Open Datasets

Free, citable statistics on how SEC filings actually move stocks — measured, not asserted.

SEC Filing Impact Monthly

Every month Wiseek scores 14,000–27,000 SEC filings for importance (1–10) and sentiment in real time. This dataset publishes what happened next: the share of filings that moved their stock, by score, sentiment, filing family and market cap — measured as excess next-session moves (event move minus the same-session median move of all actively-filing tickers, so a broad market day doesn't masquerade as filing impact).

  • Aggregates + a per-event archive. Each release ships aggregate tables for all scores 1–10, plus events_score7plus.csv — ticker-level rows (company, form, score, sentiment, next-session result) for published filings scoring ≥7, permanently archiving the rolling 90-day Filing Impact Tracker export before it forgets. Scores 1–6 appear only as aggregates.
  • Pre-registered claims — each release states at most three findings from fixed templates with minimum-sample and effect-size gates; months where a gate fails say so instead.
  • Published ~9 days after month-end, once every T+5 window has a verified close.

Latest release: August 2026 — the first canonical monthly release: 13,760 filings with a measured next-session (T+1) move across August's 21 trading sessions, 13,694 of them also measured at T+5.

Methodology (summary)

  • Event window: publication price → exact next-session close (T+1), New York trading calendar. A cumulative T+5 column is included and clearly labeled as containing T+1.
  • Excess: event move minus the same-session close-to-close median of the filer panel (tickers with ≥1 SEC filing in the 90 days ending at month-end and a real close pair — the full price universe is ~74% stale reprints and would pin the baseline at zero). The two legs differ (print→close vs close→close) and are disclosed; the daily tape series ships in meta.json.
  • Exclusions, all counted in the open: missing closes, calendar gaps, sub-10¢ stocks, ±300% cap, split windows, late price captures. One observation per ticker per day (highest score wins).
  • Integrity: scores and sentiment are point-in-time as stored at enrichment — never re-scored after outcomes are known. Statistics are withheld for any cell with fewer than 30 measured filings. Language is "moved after", never "caused by".

Full conventions mirror the live Filing Impact Tracker methodology.

License & citation

All datasets are CC BY 4.0: free for any use, including commercial, with attribution. Cite as "Wiseek Filing Impact Monthly — wiseek.ai/datasets/" with a link to the release page you used. A checksummed mirror lives on GitHub (WiseekAI/wiseek-datasets), and every release is archived with a DOI: 10.5281/zenodo.22047734. Questions or corrections: contact us — see our editorial policy.

Open data API

Every table is also served as JSON: static files, CORS-open, edge-cached, no key, no rate cards. Start at the catalog and follow the file URLs:

curl https://wiseek.ai/datasets/index.json
curl https://wiseek.ai/datasets/2026-08/impact_by_score.json

Machine-readable spec: openapi.json. Shape: {schema_version, period, rows}; checksums for every file are in each release's manifest.sha256 and the catalog. Attribution (CC BY 4.0) applies to API use too.

Releases

  • Launch Note — Jul 24 to Aug 14, 2026 — the first 16 trading sessions of measurement: 11,416 filings with clean next-session data (T+1 only).
  • August 2026 (2026-08) — first canonical monthly release: 13,760 filings with clean next-session (T+1) data across 21 trading sessions; adds the T+5 window (13,694 measured). Canonical for the days it shares with the launch note (Aug 1–14).