Gross Versus Net: What Grayscale's Mini ETH ETF Flows Reveal About Real Demand

The Headline I Didn't Trust

I keep a small ritual that has served me for twenty-five years in this industry. Whenever a flow headline arrives with a round number in it — and "over $100 million" is the classic round number — I open the underlying data before I open the press release. Round numbers are usually numbers that have been chosen for me. Chosen numbers deserve the same skepticism I once applied to token distribution schedules, and the habit has never failed me yet.

So when the wire told me that Grayscale's Ethereum Mini Trust had absorbed more than $100 million of ETH inside a twenty-day window, I did the arithmetic the headline was quietly hoping I would skip. The Mini Trust took in $103.3 million. Over the same window, its older sibling — the Grayscale Ethereum Trust, known to everyone as ETHE — released $71.6 million. Subtract one from the other and the family's actual net gain was roughly $31.7 million.

That is the whole story. Everything that follows is the argument for why that subtraction matters more than the headline, and why the number $31.7 million describes a company defending its own customers rather than a market discovering a new asset.

I want to be precise about what I am and am not claiming. This is not an accusation of fraud. It is not even an accusation of bad faith. Grayscale disclosed the numbers; a reader simply had to want to see them. What concerns me is the gap between the size of the gross figure and the size of the net figure, and how reliably that gap disappears in the retelling. In a bull market, that gap is where most of the misunderstandings live, and the people who profit from the misunderstanding are rarely the people holding the shares.

Truth over hype. Always.

A Company That Invented the Category and Then Lost the Fee War

To understand why a $31.7 million net figure should interest you more than a $103.3 million gross figure, you have to understand what Grayscale is and what has happened to it over the last three years.

Grayscale did not merely participate in the institutional adoption of crypto; it invented the on-ramp. When it launched the Bitcoin Investment Trust in 2013, there was no spot Bitcoin ETF, no regulated vehicle that a pension consultant could hold without flinching. For years, Grayscale's trusts were the only game in town for anyone who wanted crypto exposure inside a traditional brokerage account. The trust structure was closed-end: shares traded on the secondary market, creation was periodically suspended, and there was no redemption mechanism. For a long stretch, that rigidity produced a persistent premium, because demand for exposure outstripped the supply of shares. In 2020 and 2021, GBTC traded at premiums that made the trust itself a speculative instrument — a way to bet on future inflows rather than on Bitcoin itself.

Then the premium inverted, and the same structural rigidity that had manufactured the premium manufactured the discount. With no redemption mechanism, shares could only be sold into the secondary market, and as the 2022 bear market took hold, GBTC traded at a discount that at its worst approached fifty percent of net asset value. Holders who had bought the premium were now trapped inside the discount. It remains the single most expensive lesson in structural design that this industry has ever administered to its own institutional clients, and I watched it unfold with the particular frustration of someone who spends her working life explaining why structure matters before it matters.

The remedy was conversion, and the remedy required a lawsuit. Grayscale sued the SEC after the agency rejected its application to convert GBTC into a spot Bitcoin ETF, and in August 2023 the D.C. Circuit ruled that the rejection had been arbitrary and capricious — the agency had approved Bitcoin futures ETFs and could not rationally reject a spot ETF on the same record. That decision is the legal hinge on which the entire spot crypto ETF era turns. When the SEC finally approved spot Bitcoin ETFs in January 2024, and spot Ethereum ETFs that July, it did so under the shadow of a court that had already told it where the law pointed.

And then Grayscale made the mistake that defines its current position. It brought GBTC to market as an ETF carrying a 1.5% management fee, while BlackRock's iShares Bitcoin Trust launched at 0.25% and several competitors waived fees entirely for a promotional period. The market did not deliberate. It left. GBTC shed billions of dollars in the months after conversion, and BlackRock's IBIT overtook it as the largest Bitcoin ETF in the world. The lesson was brutal and simple: in a commoditized product, brand loyalty lasts exactly as long as the fee differential allows, and not one day longer. Trust is the only currency that matters, and even trust is not immune to arithmetic.

The Ethereum side repeated the pattern with almost mechanical fidelity. ETHE converted to a spot ETF in July 2024 carrying a 2.5% fee — an astonishing number in a market where the cheapest products were charging a tenth of that. Grayscale launched the Ethereum Mini Trust on the very same day, at 0.15%, as a lower-cost sibling. The structure was explicit: if you were a Grayscale client who could not bear to leave the family, you now had somewhere to go that did not cost you 2.5% a year. ETHE bled from the start, and it has not stopped bleeding.

So when the recent headline announced that the Mini Trust had absorbed $103.3 million in twenty days, the honest framing is not "Grayscale's ETH ETF is winning." The honest framing is "Grayscale's low-fee product is doing the job Grayscale designed it to do: catching the customers who are fleeing Grayscale's high-fee product." That is defense, not offense. It is a company managing its own churn. And churn, as any asset manager will tell you over a drink, is the most expensive problem in the business.

One more piece of context belongs here, because it is the piece most writers leave out. Grayscale's parent company is the Digital Currency Group, and DCG's recent history includes the 2023 bankruptcy of its lending arm, Genesis, and a subsequent restructuring. The trusts are legally separate from DCG and its other entities; Grayscale's counsel will tell you that, and they will be correct. But reputation is not a legal fiction, and neither is the incentive to keep assets inside the family when the family has been through a restructuring. I raise it not as an accusation but as a variable. When you are evaluating whether a product strategy is driven by client need or by parent-company pressure, you are entitled to know who the parent is.

Noise filtered. Signal preserved. The signal here is a fee structure in retreat, and a family of products quietly rearranging its own customers.

How an ETF Actually "Buys" ETH

The first thing to internalize is that a spot ETF does not buy ETH the way you and I buy ETH. It buys ETH through a mechanism that most coverage glosses over, and the gloss is exactly where the confusion about "inflows" begins.

An ETF operates through authorized participants, or APs. These are large financial institutions — market makers, typically — that have a contractual right to create and redeem shares directly with the fund. When demand for shares rises, the AP delivers the underlying asset (or cash) to the fund and receives newly minted shares in return. When demand falls, the AP returns shares and receives the underlying back. This creation-and-redemption loop is what keeps an ETF's market price anchored to its net asset value. Without it, you would have the closed-end trust problem all over again: a premium that becomes a discount that becomes a trap.

Here is the crucial detail. The ETFs on the market today are overwhelmingly cash-create and cash-redeem instruments, not in-kind instruments. That means when an AP creates shares, it hands the fund cash, and the fund — or its custodian — then goes and buys ETH in the market. The reverse happens on redemption: the fund sells ETH and hands cash back to the AP. In an in-kind world, the ETH would move directly between the AP and the fund, and no market sale would occur at all. In the cash world we actually have, every redemption is a potential sale into the order book, and every creation is a potential purchase.

This distinction is the difference between an outflow that pressures the price and an outflow that does not. It is also the difference between a clean reading of on-chain data and a misleading one. When a platform like Arkham shows you an address associated with an ETF receiving or sending ETH, you are watching the mechanical result of creation and redemption activity. You are not necessarily watching a fund manager make a discretionary decision to buy or sell. The address moves; the intent behind the movement is inferred.

So the first correction I want to install in the reader's mind is this: "ETF inflow" is not a synonym for "new conviction." It is a synonym for "net creation of shares." Those two things can coincide, but they are not the same thing, and in a market where a single issuer runs two products tracking the same asset at wildly different prices, they diverge almost completely.

The Arithmetic of the Headline

Now the subtraction. It is almost embarrassingly simple, which is why it is so easy to hide.

The Mini Trust absorbed $103.3 million. ETHE released $71.6 million. The combined Grayscale Ethereum complex therefore grew by approximately $31.7 million over the twenty-day window. That is the number that describes Grayscale's actual footprint on the ETH market. The $103.3 million describes one product's gross activity, and gross activity is a measure of motion, not of accumulation.

Why does this matter so much? Because gross figures are the ones that get amplified, and net figures are the ones that get buried. A headline that says "Grayscale ETH ETF pulls in $103 million" invites a reader to conclude that $103 million of new money arrived to buy ETH. A headline that says "Grayscale's ETH complex grows by $31.7 million, almost entirely through internal migration" invites a very different conclusion, one that is much closer to the truth. The first headline is a story about demand. The second is a story about fee competition. They cannot both be the main story, and only one of them is accurate.

I have spent enough years auditing token distributions to recognize the pattern. In 2017, when I was spending months inside the whitepapers of the EOS and Golem crowdsales, I learned that the number a project chooses to put at the top of its announcement is almost never the number that determines whether the token economics are sound. The headline figure is chosen for its psychological effect; the structural figure is the one that decides outcomes. The same discipline applies here. The $103.3 million is the psychological figure. The $31.7 million is the structural figure. Watch the structural figure.

There is a second layer to the arithmetic that deserves attention, and it is the one that makes the net figure even less impressive than it first appears. The twenty-day window is almost certainly a window that overlaps with the period around the product's ramp-up. Windows that begin near a launch date are contaminated by one-time events: seed capital moving into place, initial allocations being rebalanced, and the mechanical churn of getting a new share class to a functional float. A twenty-day snapshot taken during a product's early life is not a steady-state measurement. It is a photograph of a launch, and launches are noisy.

The Fee Is the Product

If you take one analytical tool away from this article, make it this one: for a commodity exposure product, the fee is the product. Everything else — the brand, the ticker, the marketing — is packaging around a single number, and that number is the expense ratio.

Run the math on the Grayscale Ethereum complex. ETHE charges roughly 2.5% per year. The Mini Trust charges 0.15%. For a holder with $100,000 in ETHE, the annual cost of the high-fee share class is about $2,500. The same $100,000 in the Mini Trust costs about $150. That is a difference of $2,350 per year, every year, for identical exposure to the same underlying asset. The two products hold the same ETH. They track the same price. The only thing that differs is how much of your return Grayscale keeps.

No rational, fee-aware investor stays in the expensive share class once the cheap one exists and is liquid enough to absorb them. This is not a matter of sentiment or market timing. It is arithmetic, and arithmetic does not negotiate. The migration from ETHE to the Mini Trust is therefore not a sign of anything except the predictable response of capital to a ten-fold difference in cost. It will continue as long as the differential exists, until either ETHE's assets are substantially drained or Grayscale adjusts the fee.

And here is where the "rotation" framing starts to look less like a neutral description and more like a soft-landing narrative. When the market describes the ETHE outflow as "internal rotation," it is choosing to emphasize that the money stayed inside the Grayscale family. That emphasis is flattering to Grayscale. But consider the counterfactual: if the Mini Trust did not exist, where would that $71.6 million have gone? Some of it would have gone to ETHE's competitors — BlackRock, Fidelity, and the other low-fee issuers — and some of it would have left the ETF complex entirely. The Mini Trust exists precisely to prevent that. It is a wall built to keep the water inside the reservoir. Calling the water "rotated" rather than "retained" is a choice, and it is Grayscale's choice to make, but it is not the only way to describe the same flow.

The uncomfortable truth is that internal rotation is Grayscale's second-best outcome. Its best outcome would have been to keep clients in ETHE at 2.5% forever. Its worst outcome would have been to lose them to BlackRock. Rotation is the middle result, the one that salvages the relationship at the cost of the fee. A company that is celebrating its middle result is a company telling you something about how the other two are going.

What the Chain Actually Shows

I want to be disciplined about what the on-chain evidence can and cannot support, because this is where crypto analysis most often overreaches.

The ETF's ETH is held by a custodian. The industry convention — and Grayscale's disclosed arrangement — points to Coinbase Custody, though the source material here does not name it explicitly, so I will treat it as a high-confidence inference rather than a fact. What the Arkham-tracked addresses show is movement into and out of that custodial structure. What they do not show, on their own, is whether a given movement represents a market purchase, a seed transfer, an internal rebalancing, or a settlement of a creation or redemption.

The analytical discipline is to treat the address flow as evidence of activity, not as evidence of intent. A large inflow to a custodial address is consistent with "the fund bought ETH on the open market" and also consistent with "an AP delivered a basket of assets to complete a creation." Both produce the same on-chain signature. Distinguishing them requires the fund's own disclosure of its creation and redemption activity, which is reported on a lag and in aggregate.

There is a further subtlety that most coverage ignores. Because these are predominantly cash-create products, the fund's actual market purchases of ETH happen at the moment of creation, executed by the custodian or an associated trading desk. The on-chain record of the custodian receiving ETH therefore reflects the completion of a process whose market impact already happened, often minutes or hours earlier. By the time the address lights up on a dashboard, the buying is done. This is why on-chain ETF flows are a better record of what occurred than a predictor of what is coming, and why treating them as a real-time demand signal is a category error.

None of this means the data is useless. It means the data needs a translator, and the translator's first job is to stop the reader from reading a custodial transfer as a conviction trade.

The Yield That Isn't There

Here is the structural cost that the headline figure never mentions, and it is the one I would want a client to understand before they buy any spot ETH ETF.

ETH is not a passive asset. It can be staked, and staking produces a yield — historically in the low single digits, varying with network conditions and the total amount staked. A holder who stakes their ETH earns something for the privilege of securing the network. A holder who keeps their ETH inside a spot ETF earns nothing, because the ETFs on the market today do not stake their holdings. This is a regulatory artifact as much as a product decision: the SEC under its previous leadership was deeply wary of staking inside exchange-traded products, and issuers stripped staking from their filings to get approved.

The result is a hidden opportunity cost that compounds silently. If staking yields roughly 3% in the current environment, then every dollar of ETH locked inside a non-staking ETF is forgoing about three cents a year. On the $31.7 million that Grayscale's complex absorbed in the window, that is close to a million dollars of forgone annual yield — and that is before you count the management fee. An ETF holder is therefore paying to hold an asset that is simultaneously earning nothing and costing something. The two drags stack.

This is not a reason to avoid ETFs categorically. For institutions with mandates that prohibit direct custody, or for investors who cannot manage keys or run validators, the ETF is a legitimate solution to a real problem, and the convenience is worth something. But the convenience has a price, and the price is the sum of the management fee plus the staking yield you never receive. When the market frames an ETF inflow as pure bullishness, it is silently pricing that convenience at zero. It is not zero. It is one of the largest hidden costs in the product.

The forward-looking question, and one I will return to at the end, is what happens if and when the regulatory stance on staking inside ETFs softens. If issuers are permitted to stake, the economics of these products change materially, and the competitive landscape rearranges itself again. Until then, every ETF is a yield-free wrapper, and the holder is paying for the privilege of earning nothing.

The Quiet Drain on DeFi

There is a second-order effect that almost no flow report mentions, and it deserves its own section because it plays out over years rather than days.

ETH that sits inside an ETF does not participate in on-chain finance. It does not get staked. It does not get supplied to a lending market. It does not get paired in an automated market maker. It does not get used as collateral. From the perspective of the Ethereum economy, ETF-held ETH is inert. It has been withdrawn from circulation not merely as a tradeable asset but as a productive one.

This matters because the on-chain economy competes for the same ETH that the ETF absorbs. Every unit of ETH that a long-term holder decides to park in a custodial ETF is a unit that is no longer available to secure the network through staking, to deepen the liquidity of decentralized exchanges, or to underwrite lending markets. The ETF is not malicious in doing this; it is simply offering a convenience that pulls capital away from a set of on-chain uses that require active participation. But the aggregate effect, if ETF holdings grow large enough, is a slow structural thinning of on-chain liquidity and staking participation.

I have written before, and I will keep writing, that liquidity fragmentation is too often a manufactured problem — a story that venture investors tell to justify a new product that fragments liquidity further. This is a rare case where the concern is genuine, and it is genuine precisely because it is not fragmentation. It is consolidation. The ETH is not being scattered across a hundred chains; it is being gathered into a handful of custodial vaults where it does nothing at all. That is a different and in some ways more serious problem, because fragmentation at least leaves the capital in play.

The practical implication for a reader is this: when you evaluate whether ETH ETF inflows are "good for Ethereum," you have to decide what you mean by good. For the price, a marginal net inflow is a marginal positive, because it removes ETH from the tradeable float. For the network, the same inflow is a marginal negative, because it removes ETH from productive use. These two verdicts are not contradictory. They are simply answers to different questions, and the industry tends to ask only the first one because it is the one that moves price.

A Custodial Bridge With a Different Threat Model

There is a comparison that has been nagging at me since the first spot ETFs launched, and this flow story gives me a reason to state it plainly.

Cross-chain bridges have been hacked for cumulative losses exceeding $2.5 billion, and the industry nonetheless depends on them to move value between environments that cannot natively talk to each other. I have spent years documenting that paradox. The bridge is the single most attacked surface in this industry, and we keep building on it because the alternative — staying confined to one chain — is worse. Every bridge is a promise that a set of validators or a custody arrangement will honor a claim on assets held elsewhere, and every so often that promise breaks.

A spot ETH ETF is a bridge in everything but name. It is a mechanism that lets capital move between the traditional financial system and the Ethereum network, and it works by holding the real asset on one side and issuing a claim on the other. The claim trades in brokerage accounts; the asset sits with a custodian. The holder never touches the underlying. If you strip away the regulatory vocabulary, this is the same architecture as a wrapped asset, and it carries the same fundamental question: do you trust the entity holding the real thing to honor the claim?

What changes is the threat model. A cross-chain bridge is attacked through code — a reentrancy bug, a compromised validator set, a misconfigured upgrade. An ETF custodian is attacked through the traditional vectors: operational failure, insider risk, regulatory seizure, or a bankruptcy that puts the assets into an estate. The ETF's custody arrangement is regulated and audited in ways that a DeFi bridge is not, and that is a real advantage. But the advantage is a difference in kind of risk, not an elimination of it. The holder has not escaped the need to trust an intermediary. The holder has simply chosen a different intermediary, one wearing a suit instead of an audit badge.

I raise this not to alarm anyone but to correct a framing. The ETF is frequently presented as the safe, institutional, grown-up way to hold ETH, implicitly contrasted with the wild and hackable world of on-chain bridges. That contrast is real but incomplete. Both are custodial claims on assets held elsewhere. Both concentrate trust in a small number of entities. The ETF simply concentrates it in entities that regulators watch closely, which is a meaningful but not infinite comfort. Trust is the only currency that matters, and the ETF does not abolish the need for it — it relocates it.

The Competitive Blind Spot

The single biggest weakness in the source material, and in most coverage of this story, is the absence of competitors. The Mini Trust's inflow and ETHE's outflow are reported in isolation, as if Grayscale were the only issuer of ETH ETFs. It is not, and the omission is not neutral.

BlackRock and Fidelity both run spot ETH ETFs, and both charge fees in the 0.25% range — roughly ten times the Mini Trust's fee, but a tenth of ETHE's. They also bring something Grayscale cannot easily replicate: vast distribution networks, advisor relationships built over decades, and a brand that a compliance department recognizes without hesitation. In a commodity product, distribution is often the decisive advantage, because the product itself is undifferentiated. If a client's advisor already uses BlackRock for twenty other things, the ETH ETF is one more line item, and the marginal cost of adding it is nearly zero. That is a moat that no fee cut fully bridges.

The Bitcoin precedent should be studied, not merely cited. When GBTC converted to an ETF, it held the largest pool of assets and the oldest brand in the category. Within months, BlackRock's IBIT had overtaken it, and Grayscale's flagship had shed a substantial share of its holdings. The mechanism was not that BlackRock offered a better product in any technical sense. It offered a cheaper one, distributed more effectively, and the market rewarded it. There is no structural reason to believe the ETH market will behave differently, and the early flow data is consistent with the same migration already underway.

Without competitor data for the same twenty-day window, we simply cannot judge whether Grayscale's $31.7 million combined net gain represents strength or weakness. If BlackRock and Fidelity together absorbed multiples of that figure, then Grayscale is losing the category even as its own low-fee product posts an impressive-looking gross number. The "rotation" story would then be revealed as a story about a company losing share while keeping a brave face. I am not asserting that this is what happened, because the data is not in front of me. I am asserting that the data is not in front of the reader either, and that a story told without competitors is a story told without a baseline.

The Single-Window Trap

One more discipline before I move to the contrarian view, and it is the discipline that separates a market brief from a market rumor: never treat one window as a trend.

Twenty days is not a trend. Twenty days is an anecdote with a timestamp. Trend requires multiple windows, ideally spanning months, and it requires the exclusion of the distortions that cluster around product launches. If the twenty-day window here overlaps with the Mini Trust's ramp-up, then a meaningful share of the $103.3 million could be seed capital finding its home, initial allocations being moved into place, or the mechanical settlement of early creations. None of those would recur in a steady-state month, and all of them would inflate the headline.

The honest way to read a launch-window flow figure is to hold it lightly. It tells you the product exists and that someone is using it. It does not tell you the rate at which it will keep being used, and the rate is the only thing that matters for a thesis. A product that absorbs $103 million in its first twenty days and then absorbs nothing for the next sixty is a product with a launch spike and no demand. A product that absorbs $30 million a month for a year is a product with a business. The headline cannot distinguish between the two, and neither can anyone who reads only one window.

This is the discipline I try to teach the younger analysts on my team, and it is the discipline that the bull market punishes hardest, because in a bull market every single data point looks like the beginning of a trend. My rule is boring and it works: before you draw a line through one point, make sure you have at least three.

The Contrarian Case: This Is a Story About Weakness, Not Strength

The consensus reading of this flow data is upbeat. Grayscale's Mini Trust is gathering assets; the ETH ETF category is alive; institutional money is interested; the rotation shows that Grayscale can retain clients. Under this reading, the story is one of a company adapting successfully to a competitive market and holding its ground.

I want to argue the opposite, and I want to argue it carefully, because a contrarian view that is merely contrarian is worthless.

The Mini Trust's success is evidence of Grayscale's weakness, not its strength. Think about what it means for a company to launch a product whose primary function is to cannibalize its own flagship. ETHE was, for years, Grayscale's Ethereum franchise, and it carried a fee that reflected the era when Grayscale had no competition. The Mini Trust exists because that era is over. It is a defensive instrument, deployed to stop customers from walking out the door, and the fact that it is working is not a sign of health — it is a sign that the walking was real and urgent. A company does not build a lifeboat unless it expects to need one.

The "rotation" narrative is the soft landing, and soft-landing narratives are always told by the party that needs the landing to be soft. If the $71.6 million leaving ETHE is rotation, Grayscale keeps the assets and loses only the fee. If it is exit, Grayscale loses the assets entirely, and the loss compounds because those assets were generating 2.5% a year. The difference between those two outcomes, for Grayscale's revenue, is enormous. So Grayscale — and the coverage that echoes its framing — has every incentive to call it rotation. The reader should note the incentive and discount accordingly.

Here is the deeper contrarian point, and it is the one I would stake the analysis on. The market treats ETF flows as a sentiment signal, as evidence of what investors believe about ETH. For a commodity ETF in a fee war, that reading is mostly wrong. The flow is a fee phenomenon, not a conviction phenomenon. Money moves between ETHE and the Mini Trust not because anyone changed their mind about Ethereum, but because a ten-fold difference in expense ratio makes the move free money. The flow tells you about the expense ratio, not about the asset. And when you strip out the fee-driven migration, what remains is a modest $31.7 million of net accumulation — real, but small, and unremarkable relative to the size of the ETH market.

If that is right, then the entire emotional charge of the headline — the sense that something big is happening, that institutions are piling in, that ETH demand is surging — is built on a fee arbitrage dressed in bullish clothing. The real signal in this data is not about ETH at all. It is about Grayscale's declining pricing power and the commoditization of a product that once had none. Truth over hype. Always, and especially when the hype is wearing an institutional suit.

What I Would Watch Next

The forward-looking question is not whether the Mini Trust keeps gathering assets. It almost certainly will, because the fee differential guarantees it, and guarantees are the most reliable predictions in finance. The question is what the combined Grayscale Ethereum complex does, and whether it grows or shrinks against a competitive baseline.

If the combined complex keeps growing — if the net across both products stays positive and accelerates — then Grayscale has found a way to defend its franchise, and the rotation story is vindicated. If the combined complex stagnates or shrinks while the Mini Trust posts impressive gross numbers, then the rotation is a slow leak disguised as a reshuffle, and the honest reading is that Grayscale is losing the ETH category the way it lost the Bitcoin one. Watch the combined figure. It is the only number that cannot be dressed up.

I would also watch the ETHE bleed rate for the shape of the curve. A decelerating bleed suggests the migration is nearing completion and Grayscale's remaining assets are the sticky ones. An accelerating bleed suggests the opposite, and would raise the probability of a further Grayscale response — a fee cut on ETHE, or a merger of the two products into a single share class. Such a move would be sensible, and it would also be an admission that the two-tier structure was always a way station rather than a destination.

Beyond Grayscale, the signals that matter are the ones the source material omits. Watch the competitor net flows, because a category is judged against its category, not in isolation. Watch the combined ETH ETF net across all issuers, because that is the true measure of whether ETH is being absorbed or merely shuffled. And watch the regulatory stance on staking inside ETFs, because if that door opens, every economic calculation in this article changes, and the products that looked expensive yesterday may look reasonable tomorrow.

What troubles me, and what I will leave the reader with, is the gap between the size of the story and the size of the fact. A $103 million headline was manufactured from a $31.7 million reality, and almost nobody noticed, because the headline was designed not to be noticed. If a gap that large can open between a number and its framing in a story as small as this one, how many other headlines in this bull market are rounding errors dressed as trends? The answer is not comfortable, and the only defense is the boring one: open the underlying data before you open the press release, subtract the outflow from the inflow, and read the net. The market will always offer you the gross figure. The signal is in the subtraction.

Noise filtered. Signal preserved.