Get the weekly note

Recomputed every trading night. 1429 funds scored across 5 categories, as of .

How the ETFIQ scores are computed

One score for each kind of fund. Each is a percentile of a single published figure inside the set of funds that figure can honestly be compared across. There is no score across the categories and there will not be one.

The arithmetic, in full

For a fund with value v in a set of n values, the score is

score = 100 * (count(x <= v) - count(x == v) / 2) / n

rounded to one decimal. Ties take the same score, which matters: 161 income funds sit within a tenth of a point of another, and ranking those against each other would be inventing a difference the data does not have. Where a score blends two figures, each is scored this way first and the two are averaged at the stated weights. Those weights are the only judgment in the whole exercise and they are printed beside every score they affect.

Why a percentile and not a rating

It has no weights to argue with
A percentile is a count. Four of the five scores here have no weights at all: they are a single published figure, ranked.
It survives the outliers a linear scale cannot
The income gap runs from -137 to +95 points. Mapping that onto 0 to 100 linearly would put nine tenths of the category inside 11 points of each other.
It can be recomputed by anyone
The figure, the set and the arithmetic are all published. The file carries every score, every component and every weight.
It moves when the set moves
This is the one property that has to be said out loud. A fund can slip two points on a day it did nothing at all, because two other funds improved. That is what a percentile is. It is why the underlying figure is printed beside every score, and why the score is never the only number on a page.

The rules that apply to all five

One score per category, none across them
A buffer fund and an income fund are measured on different questions. Averaging four answers to four questions produces a number that answers none of them.
A minimum set of 12
Below twelve funds a percentile says more about the set than about the fund, so no score is published for that category at all.
A high score is not a recommendation
Nor is a low one a warning. Every score page states what its own top and bottom mean, because for two of these scales neither end is the good end.
A fund with no figure has no score
It is not given a zero and it is not given the median. It is absent from the set, and the set size printed beside every score is the number of funds actually in it.
The score never enters another figure
Nothing on this site is ranked, filtered or weighted by a score. It is an output of the published data, never an input to it.

The five

Downside Cover Score, for buffer ETFs

The question: If the market falls into a bear market from here, how much does the buffer absorb?

The set: 293 buffer ETFs with a live outcome period. A fund outside that set has no score rather than a low one.

What it is computed from:

What it does not say: It is measured over a 20% fall because that is the conventional bear market line, so it says nothing about a 5% one, where a fund whose buffer has not yet engaged absorbs nothing at all. The upside left is printed beside it rather than blended into it, because the two are only loosely related and the funds where they disagree are the ones worth looking at.

Every fund in this category, ranked by it · How the underlying figures are computed

Return Stability Score, for income ETFs

The question: Was the payout funded by returns, or by your own capital?

The sets: 162 income ETFs writing on an index or proxy, and 61 income ETFs writing on a single company, ranked separately. A fund outside both has no score rather than a low one.

What it is computed from:

What it does not say: A fund can lead the category on cash paid and sit at the bottom of this scale, which is the whole reason the measure exists: it counts what a holder finished with, not what was paid out. Funds writing on a single company are ranked among themselves, because a gap to one stock and a gap to an index are not the same measurement.

Every fund in this category, ranked by it · How the underlying figures are computed

Off-Index Score, for thematic ETFs

The question: How much of it is not the index it is sold as an alternative to?

The set: 349 thematic ETFs with a holdings filing. A fund outside that set has no score rather than a low one.

What it is computed from:

What it does not say: 100 minus weight overlap is active share, which is Cremers and Petajisto and not an ETFIQ invention; what is ETFIQ’s is the percentile inside this desk. A fund at the top holds almost nothing the index holds at the weights the index holds it. Whether that is worth its fee is a question this site does not answer, and it says nothing about how diversified the fund is: a fund can overlap the index by nothing and hold five stocks.

Every fund in this category, ranked by it · How the underlying figures are computed

Decay Resistance Score, for leveraged ETFs

The question: Did it keep up with its own daily multiple, compounded day by day?

The sets: 391 leveraged ETFs, long, over three months, and 121 inverse ETFs over three months, ranked separately. A fund outside both has no score rather than a low one.

What it is computed from:

What it does not say: It measures a fund against its own promise, a multiple of each daily move compounded day by day, not against the other funds’ returns. A fund can score well in a quarter it lost money in. What is left in the number is the fee, the cost or income of financing, and the slippage of rebalancing daily, which is why long and inverse funds are ranked separately: an inverse fund earns interest on the collateral behind its short and sits about 4.8 points ahead of a long one before any question of how it is run.

Every fund in this category, ranked by it · How the underlying figures are computed

Coin Capture Score, for crypto ETFs

The question: How much of what holding the coin gave did a holder of the fund keep?

The sets: 28 spot crypto ETPs listed less than a year, and 24 spot crypto ETPs with a full year, ranked separately. A fund outside both has no score rather than a low one.

What it is computed from:

What it does not say: These funds all hold the same kind of thing, so the score is close to a fee ranking with the market’s pricing of the shares in it. It says nothing about the coin. Funds listed less than a year are ranked among themselves, because a tracking gap compounds and a short record has not had time to accumulate one.

Every fund in this category, ranked by it · How the underlying figures are computed

What is not computed

There is no risk score, no quality score, no rating and no composite. ETFIQ does not model what a fund will do next and does not hold a view on whether any fund is worth owning. Every score here is a position in a named set on a named published figure, and it is published with the figure beside it so that a reader can disregard the score and read the figure.

Where to next

Cite this page. ETFIQ, How the ETFIQ scores are computed, data as of Sep 12, 2026. https://etfiq.com/methodology/scores Free to use with attribution; the underlying files are at Open data.