This specification describes the data as published on Sep 18, 2026, covering 81 funds. Rebuilt every trading night.

Alternatives ETF methodology

This is the working specification: what is in it, where every figure comes from, how it is worked out, what is checked every night, and what it cannot tell you. It is written to be checkable by someone who does not trust it.

Scope

81 US listed alternatives ETFs across 8 strategies. A fund is here when its name points to an alternative strategy and the strategy section of its own latest prospectus describes one. Neither is enough alone: many funds are called hedged and hedge only their currency, and many prospectuses mention long and short positions or tail risk in passing.

A fund another desk already measures stays there. Option income funds that hedge are on the income desk, funds with a buffer and a cap on the buffer desk, and leveraged and inverse funds on the leverage desk. A fund whose name says it is an income fund stays with income too. 19 funds whose names suggest an alternative strategy and whose prospectus does not say so are not published and are reviewed by hand. CSM and HOLD are read on an editorial decision, because their names do not say what they do, and the prospectus still decides the strategy. FMF is included on an editorial decision: its name states the strategy and the prospectus does not settle it.

The strategies

ETFIQ uses Morningstar’s category names, from its October 2025 US definitions, as the framework, and prints plainer names. Morningstar licenses its own fund assignments; every assignment here is ETFIQ’s own, made from public filings. Two Morningstar categories are left out on purpose, because this site already has a desk for each: derivative income is the income desk and defined outcome is the buffer desk.

Sources

What is computed

Weekly returns

52 seven-day returns ending on one date for the fund and for the S&P 500: the earlier of the fund’s last close and SPY’s. Each week runs from the close on or before one date to the close on or before the date seven days later, with distributions reinvested. Weekly rather than daily, because a futures fund or a fund holding foreign shares prices at different times from the New York close, and daily returns would pair days that did not see the same news. The year measured today runs from Sep 19, 2025 to Sep 18, 2026.

The S&P 500

Measured through SPY with its dividends reinvested, because a holder of the index is paid its dividends and the index itself is not.

Correlation

The Pearson correlation of the fund’s 52 weekly returns with SPY’s. At 1 the two rise and fall together, at 0 there is no pattern, and below 0 the fund tended to move the other way. Stored and printed to two decimals.

Beta

The covariance of the two sets of weekly returns divided by the variance of SPY’s: how far the fund moved, on average, for each 1% move in the index. Two decimals.

A down week

A week in which SPY’s return was below zero. A week that ended exactly level is not a fall. The year to Sep 18, 2026 had 22 of them.

Down-week capture

100 times the fund’s average return in the down weeks divided by SPY’s average return in the same weeks, taken on the unrounded averages. At 100% the fund fell as much as the index, at 50% half as much, and below zero it rose. It needs at least 8 down weeks, and a fund whose year had fewer has no figure. Stored to one decimal. This is not Morningstar’s downside capture ratio, which is usually monthly, compounded and over three years, so the same fund shows a different number on the two sites.

Against T-bills

The fund’s total return minus BIL’s, a fund of one to three month Treasury bills, over the same 52 weeks, in percentage points. It sits beside the score and is not part of it: a fund can steady a portfolio by losing money slowly, and the reader should see that on the same line.

The Diversifier Score

The average of two percentiles inside one set, the alternatives ETFs with a full year of weekly returns: one on correlation and one on down-week capture, each ranked so that lower is better, weighted evenly. 63 funds are in the set today. How every score is computed.

The answer in words

Every fund page answers on the down-week capture rounded half up to a whole number, so the words and the figure beside them cannot disagree. Below zero the fund rose when the S&P 500 fell; at zero it barely moved; from 1 to 94 it fell that share as much; from 95 to 105 it fell with the index; above 105 it fell more, stated as a multiple of the index’s fall.

Comparisons

Two alternatives ETFs are compared only when they run the same strategy and neither is thinly quoted, and only over the same weeks. Each fund is also compared with the index funds it is most often held beside: SPY, VOO, IVV, QQQ, AGG, BND, TLT and GLD.

Net assets

The freshest figure that carries a date: the issuer’s own daily file or fund page where one states it, then the latest N-PORT filing, which reports a month end up to several months old. Of the 81 funds, 77 carry a figure today, 47 of them from a filing and 30 from the issuer. Each fund page says which source and the date.

Thinly quoted funds

A fund whose price sat unchanged on more than one trading day in twenty is published with the reason and is not ranked, the rule every desk here uses.

How a strategy is read

From the strategy section of the fund’s latest 485BPOS on EDGAR, found by the fund’s SEC series. A phrase that names a strategy decides it, such as managed futures, merger arbitrage or market neutral. A phrase that only describes one, such as trend following, long and short positions or a collar, counts only when the fund’s name points to the same strategy, and a trend, event or macro phrase also needs the fund to trade futures or sell short. A phrase in a sentence that denies it does not count. A prospectus naming two or more strategies is multi-strategy unless the fund’s name settles which. Every fund page links the filing its strategy was read from.

Where to next

Cite this page. ETFIQ, Alternatives ETF methodology, data as of Sep 18, 2026. https://etfiq.com/methodology/alternatives Free to use with attribution; the underlying files are at Open data.