Corporate actions scanner

corporate-actions-scanner

You hand over a ticker or watchlist and a lookback window (default 180 days). The skill pulls every 8-K filing from SEC EDGAR in the window, classifies each into a materiality bucket, cross-references Massive news for the headline, and computes the T+1 and T+5 price reactions.

Motivated by a live portfolio review that missed an ALLO public offering (87.5M shares at $2, 34% dilution) because news-scanner defaulted to a 24-hour window and the offering was 78 days old. Nothing else in the toolkit surfaced it.

When to invoke

• The operator is running a portfolio review and wants retrospective corporate-action coverage across the book
• "Why did X move" for an unexplained gap that predates the news- scanner default window
• Pre-trade hygiene on a name: is there a pending offering, spin-off, or M&A announcement in the last quarter that changes the setup
• The user says "material events", "corporate actions", "8-K scan", "offerings", "buybacks", "M&A on my book"

What you need

• MASSIVE_API_KEY for news + reactions (Stocks Basic is sufficient)
• SEC EDGAR is public and requires no key

What you get back

Layer 1 canonical JSON matching output-schema.json. Per-event: filing date, 8-K items, materiality buckets, matched news headline (when found), T+1 and T+5 close-to-close reactions.

Layer 2 rendered stream ranked by |T+5 reaction|. Header per event with ticker + date + items + buckets, headline underneath when matched, reaction line, optional URL to the source article. See references/rendering.md.

How it works

1. Fetch ticker metadata for the CIK (SEC EDGAR is CIK-indexed).
2. Pull 8-K filings from data.sec.gov/submissions/CIK{CIK}.json, filter to the lookback window, drop non-material items.
3. Fetch news window via /v2/reference/news for the same window, used to cross-reference the 8-K with a headline.
4. Pull daily aggs covering the filing dates + a T+5 buffer.
5. Match news to filing within +/- 2 days, preferring same-day and the article with the most flavor-keyword hits.
6. Detect flavor (public_offering, atm_offering, private_placement, share_repurchase, special_dividend, stock_split, spin_off, acquisition_announcement, acquisition_target, restatement) via a curated keyword list in the news title + description.
7. Compute T+1 and T+5 reactions close-to-close from the filing date.
8. Dedupe filings with the same ticker, date, and item-bucket set (SEC sometimes emits duplicate 8-Ks for the same corporate action).

Item taxonomy

Material items surfaced by default (see ITEM_TAXONOMY in the implementation): 1.01, 1.02, 1.03, 2.01, 2.03, 2.04, 2.05, 2.06, 3.01, 3.02, 3.03, 4.01, 4.02, 5.01, 5.03, 7.01, 8.01.

Excluded from default surface (routine 8-Ks that outnumber material ones ~10:1): 2.02 (earnings — earnings-drilldown handles), 5.02 (officer/director change), 5.05 (ethics code), 5.07 (vote results), 5.08 (shareholder nomination). Pass material_only=False to include these.

Endpoints used

• GET https://data.sec.gov/submissions/CIK{CIK}.json (SEC EDGAR)
• GET /v2/reference/news?ticker={T}&published_utc.gte={from} (Massive)
• GET /v2/aggs/ticker/{T}/range/1/day/{from}/{to} (Massive)
• GET /v3/reference/tickers/{T} (Massive, for CIK)

Doesn't handle (yet)

• Match ambiguity on multi-hit windows. When multiple news articles hit within the +/- 2 day window and all contain flavor keywords, the match picks the closest by date. Rare but real.
• Private-placement lag. Some material actions surface in 8-K days or weeks after they happened; the tool reports the filing date, not the effective date.