# Release-level validation (summary)

Run on the exact price and observation files of this release (price file SHA-256 `ec3e3983c1200a4eac7c4968e853c7115e86ffece5ffe1374b91413a3ff1be18`).

**Method (fixed before the first fetch).** 40 observations drawn by SHA-256 order of their key, 8 from each of five eras between 2004 and 2025. For each, the original SEC document was downloaded fresh and the filed share count and filed market value had to appear within 400 characters after the filed security name in the document text (text normalized for old HTML and plain-text layouts; market value also tried in thousands). The matching rule was revised while the check was being built to read split-digit old HTML and decimal XML values; the sample was fixed beforehand and did not change, and no revision loosened what counts as a match.

**Result.** 40 verified (both numbers found), 0 partial, 0 neither found, 0 not retrievable.

**Meaning.** The observation file faithfully reproduces what the cited filing says for the sampled rows. It does not test whether the filed values are right, and it does not validate unsampled rows. The earlier independent comparison against three other index funds is described in QUALITY_REPORT.md.

# Quality report

## Coverage (quarter-ends 2004-09-30 to 2025-09-30, 85 quarter-ends)
- Member quarter-ends (Cohort 1.1): 42,678; priced: 42,618 (99.86%); unpriced: 60.
- Prices in file: 42,618 for 959 securities; priced by two or more S&P 500 funds: 39,852 (93.51%); single fund: 660; Total Stock Market only: 18.
- Fund observations: 140,009 (by fund: ISHARES_IVV 39,393, SPY 25,105, VANGUARD_500 40,433, VANGUARD_TSM 35,078).
- Return intervals: 41,637; not evaluated: 26; by gate: ISSUER_STATEMENT_NO_ACTION 1, MARKET_MOVE_FUND_CONFIRMED 98, NOT_EVALUATED_DISTRIBUTION_NOT_PRICED 26, SHARE_ISSUANCE_OR_REDUCTION_ADJUDICATED_NO_ACTION 10, SPLIT_LIKE_BREAK_ADJUDICATED_NO_ACTION 2, SPLIT_LIKE_SHARE_CHANGE_ADJUDICATED_NO_ACTION 31, WITHIN_RANGE 41469.
- Corporate-action rows: 383; cash exits: 95; separation exceptions: 3.

## Checks run on this release (certification record)
Exact hash binding; frozen inputs verified against the freeze tree hashes; every price recomputed from its source observations (median rule); every observation's implied price recomputed from filed market value and shares; returns recomputed from prices; interval chain continuity; security ids subset of Cohort 1.1; member-quarter coverage reconciled to Cohort membership; no CUSIP column, token or word; deterministic double rebuild.

## Evidence behind the frozen data (not re-performed for this release)
Independent validation of this chain against three other index funds: 694 price pairs all matched (largest gap 0.096%); 598 return intervals none differing by 1 percentage point; 9 of 9 testable split ratios agree within tolerance. Full report in the Persistfolio repository under `independent-validation-batches-2026-10-29`.

## Known limitations
- Fund-implied prices, not exchange closes; agreement across funds is evidence at quarter-ends only, not for the paths between them.
- Three evidenced blockers: Scripps Networks 2008-09-30, Questar/QEP 2010-09-30, Navient/SLM 2014-06-30 (the parent left the index before a separation).
- Share basis after a security's chain ends is not provided.
- Cash terms of exits come from completion filings and were not re-read from merger agreements; two exits (Sabre, Altera) are valuation-timing cases.
- No dividend information is provided or implied.
- No third-party audit of the full dataset.