For the majority of time blockchains have been around, holding cryptocurrency necessarily involved jumping through hoops and doing business with a Wild West of unknown, unregulated fintech startups that specialized in providing on/off-ramps to digital assets. Some of those emerging startups grew into household names. Others spectacularly imploded in security incidents, insider fraud or compliance scandals, taking customer balances with them— Mt Gox in 2014 and more recently FTX and Celsius in 2022. With each bull market generating outsize returns far what is available from most asset classes, this pattern of blatant fraud and security negligence would pose a unique dilemma for the next group of early-adopters contemplating digital assets. On the one hand, on-boarding with one of these platforms could unlock significant returns if the asset class continues its stratospheric rise. On the other hand, it could also result in massive losses due to operational failure of the platform, as distinct from the unavoidable investment risks from the asset losing value.

This calculus was radically altered by the 2024 introduction of Bitcoin ETFs. It would not be long before similar offerings appeared for Ethereum, Solana and Ripple. Today it is possible to trade major cryptocurrencies through garden-variety brokerage accounts that most investors already have access to. (Unless of course their brokerage turns out to be an ideologue: Vanguard decided to patronize its customer base by holding out nearly two years before granting access to these “dangerous” investment options.) This poses a question: when does it make sense to hold the underlying asset directly over holding the ETF?

Variants of this existed starting with the very first, wildly successful gold ETF, GLD by State Street in 2004. Does one stack gold coins and bullions in a vault— as late-night infomercials targeted at a certain segment urge— or is holding GLD in an investment account a better route to the same outcome?

It turns out it is easier to answer this question for digital assets. There are only a handful of situations where directly holding cryptocurrency makes sense—and in those cases, it is imperative that investors capture the optionality provided by being able to operate on the blockchain. But most retail investors under most circumstances are better off seeking exposure through an ETF.

This essay is not investment advice or even personal opsec advice on appropriate safe-keeping of digital assets. Instead we posit a hypothetical investor who has already decided to hold some digital asset in their portfolio. Their decision comes down to purchasing that cryptocurrency directly through a VASP (“virtual asset service provider”) or through an ETF wrapper. This argument is also neutral on the question of self-custody versus parking funds at the VASP. Digital assets industry has come a long way since the Mt Gox implosion or even the FTX fraud. That includes making peace with the concept of regulation, accepting as the cost of mainstream acceptance and redirecting efforts to shape the exact contours of upcoming legislation instead of avoiding it altogether. Being regulated in some capacity—NYDFS BitLicense, bank-charter or even the patchwork of 50 state money-transmitter licenses— completing SOC2 audits and publishing financials are now table-stakes. Even US regulatory frameworks are starting to catch up with 2025 signing of GENIUS bill and ongoing work on CLARITY for market structure. (On the other side of the pond, the European Union was ahead of the game as usual with MiCA.) There are still material differences in risk profile between trusting one of these companies to hold funds versus taking on that responsibility for oneself, but those trade-offs are very particular to each situation. It is a function of the third-party custodian, the opsec level of the investor, their personal comfort level with attendant responsibility and even the type of asset in question, because it determines the hardware/software options suitable for an individual or enterprise to implement self-custody.

We start by focusing on the differences between the two options, focusing on additional avenues that are enabled by direct holding that are not possible with an ETF. In the discussion that follows, we only assume the investor has access to some digital asset platform for buying and selling cryptocurrency. This could be a centralized exchange where on-boarding requires KYC or a permissionless DeFi platform mediated by smart-contracts. Where custody model makes a difference, the distinction is noted.

Unequivocal advantages:

• 24/7 trading. Cryptocurrency markets operate around the clock. ETFs only trade during market hours 9:30-4PM. Additional extended trading is available to investors who opt-in, but “extended” does not mean around-the-clock. There is also much less liquidity available during those times and no guarantee the ETF is tracking the underlying asset accurately.
• Access to a much larger selection of digital assets. ETFs exist for only a handful of the blue-chip currencies, those with the largest market-capitalization. Even as issuers race to the tail (or bottom?) of the distribution to market Dogecoin ETFs, they have an uphill battle trying to keep up with the proliferation of copy-cat blockchains.
• Using cryptocurrency for payments. This can be done either by directly transferring the asset to another blockchain address or participating in more complex layer-2 solutions such as the Lightning Network for Bitcoin.
• Participating in on-chain distributed applications (dapps) such as prediction markets or lending pools.
• Commonly zero custody fees. Securing digital assets is one of the most expensive and operationally challenging aspects of running a cryptocurrency platform. Yet most exchanges offer basic omnibus custody—where all customer funds are pooled together into a single logical wallet— as a free service, in the expectation that trading fees will subsidize that cost. By contrast an ETF charges a management fee taken out of the assets every year.
• Exemption from wash-sale rules. There is a good reason why December sees a spike in trading: investors can manufacture artificial losses (to reduce tax liability) by selling positions that have declined in value and buying the identical position back. Net effect: portfolio remains the same but now there exist “losses” to offset capital gains for the same tax year. This works because bitcoin and other cryptocurrencies are currently classified as property rather than as securities in the US. That same trading pattern would not work for equities, including ETFs that invest in cryptocurrency. Extensive IRS rules around wash-sales discourage economically meaningless trading. These rules apply even across multiple accounts, such as selling in an investment account and buying back the same asset in a 401K. No such restrictions apply when holding digital assets directly.

Conditional advantages— these may or may not obtain, depending on particulars:

• Censorship resistance. True for self-custody, not necessarily true when funds are held by a centralized platform. It is increasingly common for VASPs to implement anti-money laundering (AML) including complying with the OFAC sanctioned blockchain address list. Attempting to send funds to one of these addresses will be rejected. Customers tempted to work around that by first withdrawing to a personal wallet and then routing it to a sanctioned address may be surprised to receive a brief, cryptic email from the compliance department indicating that their account is being closed.
• Seizure resistance. Same situation; only holds for self-custody. Most reputable centralized exchanges will freeze/seize customer funds in response to a law enforcement request.
• Faster access to funds. Again true for custody, may not always hold when digital assets are parked at a centralized exchange. Most banks and brokerages have risk limits on funds movement, such as maximum amounts that can be wired in one transaction. In theory a cryptocurrency platform could allow clients to send 100% assets to a personal wallet or a competing platform but this is not a given. In fact, because blockchain transfers are irreversible, these platforms have even more stringent risk controls to avoid losses for the customer. Because that attempt to withdraw 100% of funds looks awfully like an account takeover or perhaps a romance-scam from which the customer will never be able to recover. (Aside: the question of liability for such losses has never been formally legislated. Nor is there much in the way of case law because most disputes are forced into private arbitration. VASPs will always take the stance that it was the customer’s own actions which resulted in the loss, and therefore the customer is 100% responsible for losses.)

Looking at the disadvantages:

• High trading fees. VASPs routinely charge 0.50% to execute a single order— and recall investors must pay this toll in both directions buying and selling in order to realize gains in dollars. By comparison, most US investors can trade ETFs for free or at worst for nominal fees on the order of cents. As noted earlier, ETFs do charge a yearly management fee which eats into returns. But competition between providers naturally leads to fee compression. Nowhere is this illustrated as dramatically as with Bitcoin ETFs: even before the first day of trading, providers were racing to outdo each other by announcing drastic reductions in management fees, including a handful that promised to charge exactly zero fees for the first year.
• Slippage, especially for retail investors. VASPs aimed at consumers include an additional spread on top of the actual cost of the asset being purchased, especially for “buy now” type experience which abstracts away the actual order book. This is not disclosed as a trading fee and can result in execution at prices substantially differing from the prevailing market value.
• Inefficient execution. US brokerages are free to route customer orders to different execution venues (often with surprising incentives, as in the case of pay-for-order-flow practice highlighted during the 2021 GameStop incident) but they are subject to the FINRA best-execution rule. Among other things the rule calls for due diligence in finding the most favorable— to the customer— market for fulfilling the order and prohibits introducing unnecessary middle-man . VASPs are subject to no such restriction and can have private agreements with liquidity providers that results in customers getting suboptimal execution.
• Assumption of custody risk. Security risks apply regardless of whether the investor opts for self-custody or outsources that problem to the VASP. In the former case, the investor becomes responsible for key management: setting up a hardware wallet, making sure keys are backed up offline, carefully managing withdrawals to make sure funds are not sent to the wrong address. In the latter case, the customer is still on the hook for losses when there is an account takeover or they are socially-engineered by a scammer to voluntarily transfer funds to an address controlled by that crook. While it is possible to transfer equities such as ETFs to another brokerage account, this is a much more involved process, not to mention that it requires the crook to successfully onboard with that institution— a much higher bar than generating a new wallet address.
• Limitations for tax-advantaged accounts. It is difficult to hold cryptocurrency directly in an individual retirement account, such as IRA or self-employed 401K. No such constraints apply to holding an ETF.
• Inheritance complications. ETFs held at a broker have straightforward mechanism to transfer ownership to the named beneficiary. That is a mandatory consequence of the Uniform TOD Securities Registration Act in the US; it is not an optional feature for financial institutions to compete on. Only a handful of VASPs allow designating a beneficiary; most require falling back on the probate process. Self-custody makes it far more tricky. Absent advanced planning to make wallet credentials or seed-phrase backups accessible to heirs, the assets can become completely unrecoverable.
  

Given this background, the original question can be reframed this way: Under what conditions will the advantages of direct ownership outweigh the complications? The answer for the typical American investor with long-term horizon is: rarely.

Thrilling as 24/7 trading sounds, the adrenaline rush is lost on retail investors who are not day-trading or hoping to jump on the latest memecoin release at 4AM on a weekend. These investors will gravitate to bitcoin, ethereum and similar major chains that are already well-served by ETFs. ETF coverage today extends to almost 75% of the total market-capitalization of all digital assets. What remains on the margins are L1 assets with high-volatility and low liquidity, the equivalent of penny stocks.

Similarly this investor persona has no ax to grind with the financial system in general, no ideological fixation to pay for their cup of coffee with Lightning in some grand gesture of protest directed at The Man. They are unlikely to have Metamask installed in their browser and funded with mainnet ETH, ready to interact with Web3 applications.

As for censorship resistance or the fear of arbitrary asset seizure, that is extremely relevant—in a third-world banana republic without the rule of law, where the faction in power can use the financial system to exact revenge on the politically disfavored group. In fact, given that such regimes also have strict capital controls and rapidly depreciating currency due to misguided monetary policies beholden to the autocrat in charge, bitcoin checks all the boxes as an escape hatch. That narrative becomes much less relevant in America or most of Western Europe, notwithstanding alarmist rhetoric from certain quarters that would have been very familiar to Richard Hofstadter who spoke of “the paranoid style in American politics” more than 50 years ago. Unlike the prototypical banana republic, the United States does not—generally speaking— have a history of randomly confiscating assets from broad swaths of its citizenry as a routine corollary to regime change.

To be clear, there are very legitimate concerns about existing rules that grant law enforcement too much discretion, as in the guilty-until-proven-innocent model behind asset forfeiture. There is also historical precedent of the US financial system being weaponized to exact revenge on persona non grata; politically exposed individuals have found themselves in the cross-hairs of the IRS or Treasury. In fact the cryptocurrency industry itself has earned a rightful claim to the paranoid style after being singled out for debanking during Operation Chokepoint 2.0. Incidentally, self-custody would have been of no help in that well-documented attempt to jawbone an entire politically-disfavored sector: the chokepoint in question involved access to old-fashioned fiat dollar rails. SaaS vendors and employee salaries all need to be paid in dollars, not blockchain transfers. The more general point stands: neither political activists championing controversial causes nor employees of cryptocurrency startups are representative of the “average investor” contemplating a foray into digital assets. The specter of Big Government randomly seizing the family nest egg is not a that resonates for most investors in Western countries. (It is also a logically inconsistent threat model: if a government has lost all respect for private property rights, surely they can also seize land, housing and other tangible assets that can not be spirited away to the blockchain realm.)

That leaves a narrow but highly defensible set of circumstances for the blockchain version of amassing gold bullions:

• Residing in jurisdictions without the rule of law, where arbitrary capital controls and extra-judiciary asset-seizure is a realistic threat.
• Investing in high-volatility, low-liquidity tail-assets for which no ETF exists or is likely to exist before the initial hype-and-crash dynamic common to those assets has already played out.
• Transacting on-chain, for example making peer-to-peer payments or engaging with Ethereum dapps.
• Frequent trading including off-market hours, weekends and holidays.
• Complex tax situations when loss-harvesting from declining positions— the consolation prize when number did not go up— is important.

CP