Project 4 Source Feasibility
Sources actually read in the historical work
The SEC Financial Statement Data Sets readme was obtained on 2026-10-06 and its SUB, TAG, and NUM field definitions read. The copy belongs to historical independent evidence not bundled with the offline reader; that phrase is an identity note, not a path.
| Recorded fact | Location | PIT implication |
|---|---|---|
| EX-101 XBRL attachments cover forms 10-K,10-K/A,10-Q,10-Q/A and others; filing dates begin 2009-04-15 | §2 | No pre-2009 coverage; XBRL filers only |
| Values are as filed, uncorrected, including amendments and duplicates | §2–3 | Preserve restatement versions and select with accepted |
SUB has one row per submission, key adsh; cik, sic at filing, fye, form, period, fy, fp, filed, accepted, prevrpt; acceptance after 5:30 pm EST counts as next-business-day filing |
§5.1 | accepted is an availability candidate; period maps to period-end; prevrpt marks amended submissions; filing-specific sic may supply then-known labels |
NUM key: adsh, tag, version, coreg, ddate, qtrs, uom; value is unscaled; qtrs=0 means an instant value |
§5.3 | Annual flows need qtrs=4, balances qtrs=0; verify units |
| Quarterly publication; filings accepted after 5:30 pm EST on the quarter's final business day go to the next quarterly release | §1 | Ingestion cannot precede publication. Quarterly-package ingestion may be much later than accepted |
Limits of that reading. The readme did not explicitly specify the accepted time zone; it mentioned EST in explanatory text. Verify before converting to UTC. Mappings from XBRL tags to net income, assets, equity, revenue, COGS, SG&A, interest expense, and operating cash flow, tag coverage, custom-tag proportions, and TAG values were not read or verified. No quarterly package was downloaded. Candidate tags below are an unverified checklist.
Other sources: not obtained, rights/conventions unchecked
| Need | Candidate | Status | Note |
|---|---|---|---|
| Security monthly/daily total and delisting returns | Academic/commercial price database or exchange | Not obtained | No vendor-license/adjustment check; user chooses and supplies |
| Cap and historical shares outstanding | Same | Not obtained | Filing share capital is not arbitrary-date shares outstanding |
| Historical universe/constituent changes | Same | Not obtained | Required for real as-of membership; constructed layer demonstrates rules only |
| Historical industry classifications | Filing-specific SEC sic, commercial classification |
Partial candidate | SIC differs from GICS; match research design |
| Borrow availability/rates and transaction costs | Broker | Not obtained | Outside this signal layer |
| Published aggregate factors | Frozen French snapshot | Available but unsuitable | Aggregate spread returns cannot support security-level signals |
Implication
A real-data Project 4 cannot run before prices, cap, and historical membership are supplied. SEC data sets are a candidate for post-2009 as-filed fundamentals and acceptance times, subject to tag mapping, accepted time-zone, and annual-value-selection checks. The preparation layer therefore demonstrates rules with constructed data and makes no real-company claims.
Candidate XBRL tags: not verified
Net income, assets, equity, revenue, COGS, SG&A, interest expense, and operating cash flow each have common and custom tags; companies may change tags over time. Read TAG and several actual filings before producing mappings and coverage tables.
This is the historical 2026-10-06 reading scope. The language revision did not revisit SEC online, download quarterly packages, or verify tag mappings. Source feasibility does not establish data authorization.