Every US ETF’s Full Holdings And Operational Census Is Public, Machine-readable SEC Data (N-PORT + N-CEN) And Underused

Sharing a data source that’s surprisingly underused for fund analysis: the SEC’s N-PORT and N-CEN filings on EDGAR.

– N-PORT (quarterly, structured XML): every fund’s complete position list with weights, share counts, CUSIP/ISIN, country of domicile, ASC 820 fair-value level, monthly returns, and monthly creation/redemption flows.
– N-CEN (annual, structured XML): tracking difference vs benchmark (gross AND net of fees), securities-lending activity, in-kind creation/redemption percentages, per-broker commissions, and the full service-provider roster.

What you can pull out without any paid vendor:
– Index-fund tracking split into replication vs cost. VOO 2025 was -0.4 bps vs the S&P 500 gross of fees, -16.9 bps net.
– True per-CUSIP overlap between funds. SPY vs VOO is 476 shared holdings, ~97% by weight.
– Issuer-domicile reality checks. SPY is ~97% US, ~3% Ireland/Switzerland/Bermuda/Netherlands.

Gotchas: positions are keyed on CUSIP (not ticker), so you need a CUSIP-to-ticker map to join to anything else; unit investment trusts (like SPY) file lighter N-CEN sections than open-end funds (like VOO), so some fields are legitimately empty; and the public lag is ~60 days after quarter-end.

The StockFit API does the XML parsing and CUSIP resolution if you don’t want to build it yourself.

Not financial advice, just pointing at the filings.