# Thematic ETF methodology

> How ETFIQ computes every figure for 415 thematic ETFs as of Oct 21, 2026: scope, sources, nightly checks and what it cannot tell you.

Data as of Oct 21, 2026. Rebuilt every trading night.

This specification describes the data as published on Oct 21, 2026 , covering 415 funds. Rebuilt every trading night.

## Thematic ETF methodology

What is covered, where each figure comes from, how it is worked out, and what it cannot tell you. Check any of it yourself.

415 Funds covered

Oct 21, 2026 Figures as of

Nightly Rebuilt from the raw sources

### Scope

415 thematic ETFs, of which 391 have filed a public holdings report. A fund goes in a theme by what it holds, not by its name. The themes are ETFIQ’s own editorial grouping. Nothing else here is editorial.

### Sources

| Source | What it provides | Read | Spec |
| --- | --- | --- | --- |
| SEC Form N-PORT | Every security a fund reported holding, with weights, as filed | On filing | document |
| Issuer daily holdings files | Same-day holdings where an issuer publishes them, preferred over a stale filing | Every trading night | · |
| Tiingo end-of-day | Returns and drawdown | Every trading night | document |
| Index constituent books | The S&P 500 and Nasdaq-100 books that overlap is measured against, taken from IVV and QQQM filings | On filing | · |

### Definitions and method

Say a fund holds securities with weights w i summing to 1, and an index holds weights v i summing to 1, matched security by security.

Weight overlap (ETFIQ) The weight two portfolios share in the same securities: Σ i min( w i , v i ), as a percentage. Active share (ETFIQ) 100 minus weight overlap. This is the part of the fund that differs from the index, and the part the fee pays for. Already in the index (ETFIQ) The fund’s total weight in securities the index holds at all, whatever their index weight. This differs from overlap. A fund made entirely of S&P 500 names can still have high active share, because it weights them differently. Security matching Securities match on an internal identifier, then ticker, then name. Each holding matches at most one on the other side, so two share classes of one issuer are not double-counted. ETFIQ books show an ETFIQ id, not the CUSIP, because CUSIPs are licensed. Cost of index exposure (ETFIQ) The fund’s expense ratio divided by its active share. This is the fee per unit of exposure that differs from the index. A cheap fund that mostly copies the index can be expensive by this measure. The headline fee hides that.

### Invariants

Overlap is easy to get wrong without looking wrong. Every night ETFIQ tests four things that must always hold:

- A fund against itself is 100% overlap and 0% active share.

- Overlap is symmetric: overlap( a , b ) equals overlap( b , a ).

- Active share equals 100 minus overlap.

- Known answers hold: SPY and IVV against the S&P 500 have 0% active share, and a fund with no index constituents is 0% in-index.

These tests exist because a real bug once passed every other check. Overlap used a truncated book while the holdings count showed the full filing, so a fund had measurable active share against itself. A plausible-looking number would not catch that. These tests do.

### What this cannot tell you

- N-PORT filings lag. A filed book is as of its period end, which can be months old. The filing date sits beside every figure taken from it.

- Net assets come from the freshest dated source: the issuer’s daily file, then its product page, then the latest filing. Each figure names its source and date. A filing covers the whole registered fund. If the ETF is one share class beside mutual fund classes, the filed figure covers all of them and is larger than the ETF alone.

- Overlap counts securities, not economic exposure. Two funds holding different share classes, or a swap on the same underlying, will not show as overlapping.

- Theme assignment is editorial. The holdings are not.

### How to check any of this

Every published figure is free as JSON on the open data page . Use ETFIQ's figures with attribution and the as-of date. Issuer data stays under its owner's terms. Independent code recomputes every number nightly from the raw sources, and the statistics page shows the result. Errors are listed on the corrections page .

A figure marked ETFIQ is one ETFIQ computed. Every other figure is the issuer’s or the exchange’s, and is labeled that way.

---

Source: ETFIQ, independent ETF data. Recomputed nightly from public filings and prices.
Cite: ETFIQ, Thematic ETF methodology, data as of Oct 21, 2026. https://etfiq.com/methodology/themes
Licence: https://etfiq.com/license/ Figures ETFIQ computes are free to use with attribution: ETFIQ (etfiq.com). Issuer data stays under its owner's terms.
HTML: https://etfiq.com/methodology/themes