As of Oct 22, 2026.

How the IQ Liquidity Score is computed

181 of 4,805 ETFs have an IQ Liquidity Score, as of Oct 22, 2026.

What the score measures

The IQ Liquidity Score asks one question. Can you buy or sell this fund at a fair price, at your size, today and on a bad day?

It runs from 0 to 100, and 100 is best. The scale is fixed, so a score moves only when the fund's own figures move.

Fund size and the ETF's own trading volume are left out. A small fund with liquid holdings can score high.

Each fund also shows its percentile among the scored funds in its category. Bond, international and crypto funds are naturally wider than US large-cap funds.

It measures trading conditions, not the fund. It is not a rating or a recommendation.

The six parts and their weights

Each part is scored 0 to 100. The score is their weighted average over the parts measured for the fund, with the weights rescaled to those parts.

The anchors

Each figure scores 100 at its good anchor and 0 at its bad one, on a log scale between them. A spread of 0.02% against 0.04% matters as much as 1% against 2%.

FigureScores 100 atScores 0 atMedian spread0.02%2%Daily capacity$1bn$1mMedian premium or discount0.05%2%95th percentile premium or discount0.25%5%Bad-day premium, times the usual1 times4 timesDepth at the quote$1m$5,000

Primary market points

Authorized participants are the dealers who create and redeem a fund's shares. An active one created or redeemed some in the fund's latest N-CEN year.

With two or more active, less 15 points if one did more than 80% of the dollars created and redeemed.

Less 20 points if a fund older than a year had no creation or redemption in the 12 months to its last month held. The floor is 0.

Active dealersPoints001252453 to 4655 to 98510 or more100

When a fund is not scored

Cost to trade and price fidelity are both required. Without both, the fund shows "Not scored".

Price fidelity needs at least 20 days of closing premium.

A fund under 60 trading days old shows "Too new to score", with its parts so far.

A fund under a closing notice keeps its figures but has no tier, and carries a "Closing" flag.

A part that cannot be measured says "not measured". It is never guessed or filled with a category average.

The page always says how many of the six parts were measured.

The tiers

ScoreTierFor a retail order85 to 100Trades like a giantCost is a rounding error at normal sizes.70 to 84LiquidCheap to trade with a limit order at most times.55 to 69Liquid with careUse limits, avoid the open, watch the premium.35 to 54ThinCosts are real. Patience and limits matter.0 to 34Hard to tradeExpect a wide spread or a price far from value.

The flags

Small but liquid: net assets under $100m, or median volume under 20,000 shares, and a score of 70 or more.

Big but costly: net assets of $1bn or more, and a score under 55.

Price can drift from value: a median premium or discount above 0.5%, or a 95th percentile above 2%.

One dealer: one active authorized participant, or one doing more than 80% of creations and redemptions.

Closing: the fund's issuer has filed a notice naming its last trading day.

Cost for your order

At $1,000 and $10,000 the one-way cost is half the spread.

At $100,000 and $1m it adds a market impact term that grows with the square root of the order against capacity. Those two are labelled estimates until the term is calibrated on trade data.

Full cost to own, at $10,000: the fee for each year held, plus one round trip at the spread, plus the median premium or discount.

A fair limit: the midpoint plus half the typical spread, rounded up to a whole cent.

The formulas

Lower is better (spread, premiums, bad-day multiple): s = 100 x clamp((ln bad - ln x) / (ln bad - ln good), 0, 1).

Higher is better (capacity, depth): s = 100 x clamp((ln x - ln bad) / (ln good - ln bad), 0, 1).

Capacity: C95 = the minimum, over the holdings making up 95% of the weight, of 10% of each holding's 20-day median dollar volume divided by its weight. Cash is left out.

Price fidelity = the average of the median premium score and the 95th percentile score.

Bad-day multiple = the median premium on the bad days over the median on other days, each floored at 0.05% first.

Cost at size Q = spread / 2 + k x daily volatility x the square root of (Q / C95), with k = 1 until calibrated.

Version

Method v0.1, frozen on 2026-10-20. Any change to an anchor or weight is a new version, published here.

v0.1, 2026-10-20: Bad-day behaviour anchors from 1.5 to 10 times the usual to 1 to 4 times, because Phase 1 measures the premium, not the spread.

Every file is open: trading-cost.json, trading-capacity.json, trading-score.json.

Where to next

ETF trading costs and liquidity

Every method

Standards and sources

The score measures trading conditions, not the fund. ETFIQ makes no recommendations.