Coverage of the Russell 3000 by the IWV proxy (L5, cluster I)
Prepared 2026-11-05 for the Stage D revision (items FIRC0001-ADVE-09, FIRC0001-THEO-07). Machine-readable copy: results/russell_coverage.json. All figures were read from the public documents listed below; none was taken from a search summary.
Sources read
| document | date of the figure | figure | URL | SHA-256 of the file read |
|---|---|---|---|---|
| FTSE Russell, Russell 3000 Index Factsheet (4 pages) | index characteristics as of September 30, 2026 | Number of Holdings 2,996 (Russell 3000E: 3,432) | https://research.ftserussell.com/Analytics/FactSheets/Home/DownloadSingleIssue?issueName=US3000USD&isManual=True | 281e92643781cf19a3f4bc6a670e0719aa5ef1347d1071300491274aed0436e4 |
| iShares, iShares Russell 3000 ETF (IWV) product page | as of November 4, 2026 | Number of Holdings 2,588; benchmark "Russell 3000 Index" | https://www.ishares.com/us/products/239714/ishares-russell-3000-etf | (HTML page, read 2026-11-05) |
| iShares, IWV Summary Prospectus (incorporating the Prospectus and SAI dated July 31, 2026) | — | "BFA uses a representative sampling indexing strategy to manage the Fund" | https://www.ishares.com/us/literature/summary-prospectus/sp-ishares-russell-3000-etf-3-31.pdf | (PDF, 8 pages, read 2026-11-05) |
| FTSE Russell, Russell US Equity Indexes: Construction and Methodology, v7.2, August 2026 (50 pages) | — | methodology document now cited by ftserussell2026 |
https://www.lseg.com/content/dam/ftse-russell/en_us/documents/ground-rules/russell-us-indexes-construction-and-methodology.pdf | 26d2d051beb27a6a808af7c17a2e2871cc5388e669bedbb42cbc4d0d3ce47c76 |
No factsheet dated October 31, 2026 existed at retrieval time; the September 30, 2026 issue is the nearest to the holdings date of the build (October 23, 2026). The factsheet states that the index is reconstituted semi-annually with quarterly IPO additions, so both dates fall inside the same reconstitution window.
Coverage ratio
numerator (from data/stats.json, universe) |
value | / 2,996 |
|---|---|---|
| distinct SEC CIKs in the proxy (share classes collapsed) | 2,555 | 85.3 % |
| equity lines mapped to a CIK | 2,573 | 85.9 % |
| equity lines with a ticker, as reported by the build log (file not archived) | 2,587 | 86.3 % |
Formula: 100 * numerator / index_constituents. The headline figure for the paper is the first row (issuers are the unit of the universe).
What the ratio does and does not say
- Units differ. The index figure counts securities; the proxy counts issuers after collapsing share classes (2,587 → 2,555). The issuer-level coverage is therefore higher than 85.3 % by the number of multi-class constituents, which cannot be known without the licensed constituent list.
- Dates differ (2026-09-30, 2026-10-23, 2026-11-04). The fund page count on November 4 (2,588) is consistent with the 2,587 equity lines of October 23.
- Sampling. The prospectus states representative sampling: the fund need not hold every index member, and which members are omitted is not public. Under sampling the omitted names are expected to be the least liquid, i.e. the smallest issuers — exactly the extension beyond the S&P 500 — so the proxy's coverage of the small-cap end is likely lower than 85 %. This is a qualitative statement; no number can be attached to it from public documents.
- The gap is at most 2,996 − 2,555 = 441 securities (14.7 % of the index count), an upper bound on omitted issuers.
Stage E targets (prose owned by Stage E)
- Title / abstract / §1: "Russell 3000" → "Russell 3000 proxy (iShares IWV holdings)" where the estimand is meant; the macros
\SPIDrussellIndexConstituents(2,996) and\SPIDrussellProxyCoveragePct(85.3) are generated bycode/make_tables.pyfromresults/russell_coverage.json. - §3.1 (Estimand): add one sentence — the index listed 2,996 securities on September 30, 2026 (FTSE Russell factsheet); the IWV holdings of October 23, 2026 map to 2,555 SEC registrants, 85.3 % of that count; the fund uses representative sampling (Summary Prospectus, Prospectus dated July 31, 2026), so the omitted names are not identifiable from public documents.
- §3.2 / Table 1 note: cite
ftserussell3000factsheet2026andiwvholdings2026(both inreferences.bib), and state that the 2,587 line count is reported by the build log because the holdings file was not archived. docs/universe_definition.mdandRUSSELL3000-EXTENSION.md: align the unmatched count on the 7 names ofstats.json(N10 supplies the per-name reason codes).