US-listed spot ether products are reportable holdings, which means that every quarter a slice of their shareholder base becomes visible in institutional filings. The headline that gets written from this is always a count of holders, or a total value held, and neither number answers the question that matters for anyone holding the underlying asset.
The question is composition. A dollar of ETHA held inside a wealth adviser's model allocation and a dollar held by a multi-strategy fund are the same dollar in the total and behave in opposite ways under stress. The first rebalances into weakness because the model has a target weight. The second is a position with a stop somewhere. Aggregate them and you have a number that tells you nothing about how the holder base responds to a drawdown, which is the only thing a holder base is useful for knowing.
The four products and what their filings can and cannot show
Work across ETHA, FETH, EZET and ETHW together rather than one at a time. The products differ in sponsor and fee, and holders sort themselves across them for reasons that include platform availability and existing relationships rather than any view on ether. Analysing one in isolation confuses a product-selection effect with a positioning change, and a holder rotating from one wrapper to another will read as an exit and an entry in two separate analyses.
Be clear about coverage before drawing conclusions. Quarterly filings only capture managers above the reporting threshold, holding as of a quarter-end date, filed up to forty-five days later. Direct retail ownership does not appear. Ownership through vehicles that do not file does not appear. So the visible holder base is a subset, and it is a subset biased toward exactly the intermediated professional money you want to study, which is convenient but needs stating whenever a percentage gets quoted.
The share counts are the useful field, not the values, because values move with the price of ether and share counts only move when somebody trades. A holder base that looks like it grew thirty percent in value while share counts fell did not grow.
Typing the holders, and the three roles that are not allocations
Classify each holder from its own filing rather than from its name, using the properties of the whole book. Position count, top ten concentration, share of the book held in fund wrappers, presence of option lines, and quarter-over-quarter turnover are enough to separate the cohorts that matter here.
Advisory and allocator books show large position counts, heavy wrapper usage, low concentration, and small changes spread evenly across many lines. A crypto product appearing in that kind of book at a small weight, alongside other broad exposures, is an allocation decision, and allocation decisions move slowly in both directions.
Discretionary trading books show fewer positions, higher concentration, real turnover, and option lines. The same product in that kind of book is a position, and positions have horizons measured in weeks.
Three holder roles need to be pulled out separately before either cohort is measured, because all three produce holdings that mean nothing directionally.
- Market makers and authorised participants. Their holdings are inventory arising from the creation and redemption process. Inventory moves with flow, not with view, and it will make a cohort look decisive when it is doing plumbing.
- Brokerage and custody entities reporting positions held for others. The economic owner is somebody else entirely.
- Fund-of-fund and model-delivery vehicles that hold the product on behalf of downstream advisers, which double-counts the same client money if you also capture the advisers.
What sticky looks like, measured rather than asserted
Stickiness is a property you can measure directly once the cohorts are defined, and it is worth doing rather than assuming.
The cleanest measure is survival. Take the set of holders present in a given quarter and compute what fraction are still present two, three and four quarters later, cohort by cohort. Do it on share counts held flat or higher, so that a holder who cut ninety percent does not count as retained.
The second measure is behaviour through a drawdown. Find the quarters where ether fell materially and look at what each cohort did with share counts. Allocator books that rebalance to a target weight will mechanically add shares when the price falls, because holding the weight requires it. Trading books will cut. That divergence is the actual definition of sticky money, and once you have measured it for these products you have a coefficient you can apply rather than an intuition.
The third is sizing. Position weight within the holder's own book separates a token allocation from a real bet. A product at a small fraction of a large advisory book is durable but also inconsequential to that manager, which means it will not be defended if the platform changes its model. A product at a meaningful weight in a concentrated book is fragile in the short run and consequential.
Why a long line is often not a long
This is the failure that makes naive holder analysis of crypto products worse than useless, and it deserves stating plainly.
Filings cover long positions. They do not cover short positions, futures, swaps or borrowings. A fund running a cash-and-carry structure, long the spot product and short listed futures to harvest a basis, files a long line that is indistinguishable from an outright long. Their economic exposure to ether may be close to zero. When the basis compresses they will unwind, and the unwind will look in the data like conviction selling by a sophisticated holder.
Several tells help without being conclusive. The position appearing and disappearing on a rhythm that tracks futures expiries rather than price. Simultaneous option lines on related instruments. A holder whose book is broadly full of arbitrage-shaped structures. And the crudest and most useful one, a holder type that has no business making a directional crypto allocation showing a suspiciously round position.
The honest conclusion is that hedge fund cohort figures for these products should carry an explicit caveat that some unknown share is basis positioning rather than a view. Publishing the cohort split without that caveat overstates professional conviction, and it overstates it precisely in the cohort that gets quoted most.
Using the split without over-reading it
What this analysis produces is a description of the holder base, refreshed four times a year, with a lag. It does not produce a trade, and treating a quarterly composition shift as a timing signal is a misuse, since by the time you can see it the composition has already changed again.
What it legitimately supports is an assumption about downside behaviour. A holder base weighted toward allocator money implies mechanical demand into weakness and slower supply on the way out. A base weighted toward trading books implies the opposite, and implies that a drawdown will be amplified by holders who share your reasons for owning it. That assumption feeds position sizing and stress scenarios, not entries.
It also supports a slower question worth tracking across years rather than quarters, which is whether these products are actually being adopted into allocation frameworks or are being rented by trading desks. Cohort survival across a full cycle answers that, and nothing published on a filing day does.