BOXX ETF Explained: Box Spreads, Tax Advantages, and the Tradeoffs

There is a $13.8 billion fund that holds no bonds, pays no dividends, and has made exactly one distribution in nearly four years of operation. It is designed to earn roughly what Treasury bills earn and to let that return accrue in the share price rather than arrive as taxable interest.

BOXX, the Alpha Architect 1-3 Month Box ETF, is the most interesting tax structure in retail investing right now, and the argument about it has been running since 2024 between people who know the statute extremely well. Three separate questions get compressed into one, and separating them is most of the work:

• Are box spreads economically sound? Essentially nobody disputes this.
• Is BOXX a better deal than SGOV after fees and taxes? Entirely dependent on your bracket, your state, and how long you hold.
• Will the intended tax treatment survive IRS scrutiny? Genuinely unresolved, with credentialed people disagreeing in public.

An answer to any one of these does not settle the other two. Most coverage treats skepticism about the third as if it disposed of the first, or confidence in the first as if it disposed of the third.

What BOXX holds

A box spread combines four options at two strikes with the same expiration, and the four payoffs cancel the underlying index entirely, leaving a fixed amount due at expiration. The algebra is worked through in the guide on borrowing with box spreads, which covers the same instrument from the other side. BOXX takes the lending side: it pays cash now for a known amount later.

The prospectus draws the comparison itself: “Buying (or selling) a Box Spread is similar to buying (or selling) a zero-coupon bond… The maturity value of a zero-coupon bond is comparable to the difference in the strike prices of the Box Spread.”¹

Two design choices matter. The fund keeps the weighted average maturity of its boxes under 90 days, and it uses European-style options, which “may not be terminated or assigned in advance of the option’s expiration date.” The prospectus explains why: it “ensures that none of the synthetic positions created using the Box Spread will be forcibly closed prior to the Box Spread’s maturity.”¹ Early assignment on one leg is the one failure mode that has destroyed a retail box in practice, and European exercise removes it.

As of mid-August 2026 the fund held $13.77 billion, charged 0.1949% net after a waiver running to at least February 1, 2027, and quoted a 30-day median bid/ask spread of 0.01%. Its average position sat 34 days from expiration.²

Why the fund holds SPY and QQQ options

The audited schedule of investments at September 30, 2025 shows the option portfolio consisting entirely of FLEX options on the SPDR S&P 500 ETF Trust and the Invesco QQQ Trust.³ Today’s live holdings are the same, and every option ticker carries the prefix the fund page identifies as denoting European exercise.² Meanwhile the prospectus discussion of §1256 and its 60/40 treatment is written entirely around SPX options.

Options on an ETF are options on shares of stock, so they are equity options rather than the nonequity options §1256 reaches. A fund holding SPX boxes would be marked to market every December 31 and forced to recognize gain whether or not it wanted to distribute anything. Holding ETF options instead avoids that. The fund does not explain the choice, and the prospectus reserves the right to use other underliers “when Arin has determined that doing so would provide the Fund with better risk and return or tax characteristics,”¹ so read the connection as inference rather than as a stated rationale. It is consistent with a structure built to avoid forced recognition.

Why an ETF instead of doing it yourself

You can buy boxes directly. Doing so means selecting strikes and expirations, filling four legs without legging in, holding margin approval, monitoring assignment, valuing the position, and tax reporting that depends on which contracts you used, with SPX boxes running through Form 6781. One ticker replaces all of that, and Alpha Architect describes the fund’s purpose as democratizing access to a market that was institutional.⁴

The strategy is not new. Alpha Architect’s own materials point to Billingsley and Chance publishing on box-spread efficiency in 1985. When BOXX launched in early 2023, Rick Ferri wrote on the Bogleheads forum that box spreads “have been around for decades,” that counterparty risk was “slight… but minimal,” and that the strategy came into favor after the 1987 crash because investors had capital losses and fell out of favor when rates hit zero. He then named the use case that turned out to matter most:

If you have a sizable capital loss carryforward from 2022 with no gains in sight to offset those losses against, you could use this strategy to create about a ~ 4% tax-free yield because the ~ 4% gain would be offset by tax-loss carry forward.

That was February 2023, a year before the tax debate started, and it identified the same edge case that Elm Wealth would later single out as the clearest exception to its own skeptical conclusion.

Two different tax benefits

The pitch is usually described as one advantage. It is two, and they fail independently.

Deferral. SGOV receives Treasury interest and distributes it monthly, so you owe tax every year whether you spent the money or not. BOXX has almost nothing to distribute, and like every ETF it can hand appreciated positions to authorized participants in kind without recognizing gain. The prospectus cites the provision by name: under §852(b)(6) the fund “does not recognize any built-in gain in such appreciated property.”¹ That mechanism is covered in the guide on what can happen to a capital gain, and it is ordinary ETF machinery rather than anything exotic.

Character. Hold the shares more than a year and sell, and the appreciation is intended to be long-term capital gain. That would replace interest taxed at up to 37% with gain taxed at up to 20%. Seventeen points is the widest that spread gets, and it is what the character half of the pitch is worth when it works.

Deferral may hold up better under a challenge aimed at shareholders, since recharacterizing the gain on sale would change its rate without restoring annual taxation along the way. It is not independent of legal risk either. The fund-level theory discussed below goes at §852(b)(6) directly, and the prospectus contemplates the fund failing to qualify as a RIC, paying a deficiency dividend, and having its distributions recharacterized. Both halves rest on the law as much as on portfolio management.

Three layers that get conflated

A chart circulates showing box spreads taxed 60% long-term and 40% short-term. It is accurate about one thing and misleading about BOXX, because three separate taxpayers are involved.

The 60/40 chart describes the top row. It is the wrong frame for the bottom row, which is what a BOXX shareholder experiences and which turns on the ordinary holding-period rule rather than on §1256.

What the distribution record shows

BOXX is often described as a fund that never distributes. The record is better than that description and also not quite that clean.

Three of four periods produced nothing. One produced a taxable event that shareholders did not choose. The audited financials also show net investment income per share negative in every fiscal year, running at a ratio of −0.1938% of average net assets in fiscal 2025,¹ which is why the fund publishes no SEC 30-day yield: there is essentially no accounting income to report. So describe BOXX as designed to minimize taxable distributions, and leave room for the year when it does not manage it.

The federal arithmetic

Everything below assumes a 4.00% gross return for both funds. That is a modeling choice rather than a forecast, and it is deliberate: BOXX publishes an average yield to option expiration that explicitly excludes fees, while SGOV publishes an SEC yield that is net of them. Subtracting one from the other double-counts one fund’s expenses and ignores the other’s. Holding the gross return equal isolates what is being compared, which is fees and taxes.

Net of expenses, that leaves BOXX at 3.8051% and SGOV at 3.9100%. SGOV starts ahead. The 2026 rates that follow are from Revenue Procedure 2025-32, and the top ordinary rate is still 37% because the seven-rate structure was made permanent in July 2025 rather than sunsetting back to 39.6%.⁵

The table is not monotonic in income, which is the part that surprises people. The 12% ordinary bracket paired with a 0% long-term rate produces a wider edge than the 22%/15% pairing, because the character spread is twelve points rather than seven. BOXX is not exclusively a high-earner strategy on the federal math alone. That row carries a caveat, though: the 0% long-term bracket runs only to $98,900 of taxable income filing jointly, and realizing a gain consumes that space.

The 3.8% net investment income tax appears on both sides once modified AGI passes $250,000 filing jointly or $200,000 single, because it reaches interest and capital gains alike. It shrinks both returns without creating the gap. Its thresholds have never been indexed for inflation, so more households cross them each year.

Where state taxes come in

Federal law exempts interest on U.S. government obligations from state taxation. The exemption is written narrowly. 31 U.S.C. §3124(a) reaches “the obligation, the interest on the obligation, or both,” and subsection (b) sends “the tax treatment of gain and loss from the disposition of those obligations” to the Internal Revenue Code instead.⁶ State definitions follow suit: California and Connecticut both define the exempt dividend as one “other than a capital gain dividend.”

So SGOV passes through its Treasury-source income free of state tax, 95.14% of its distribution for tax year 2025,⁷ while BOXX gains are exposed in full. Run the same six situations across a range of state rates and the picture changes materially. Positive numbers favor BOXX.

Two readings of that table cut against the simple versions of both arguments. A state income tax does not automatically settle the question for SGOV: at the widest federal spreads, BOXX survives even a 13.3% California rate by roughly ten basis points. And a high federal bracket does not automatically settle it for BOXX: at 22%/15% a state rate above about 5% is enough to erase the whole thing.

A note on the top-right cell. Pairing a 37% federal rate with a 13.3% California rate describes a specific couple, since California reaches 13.3% only above roughly $1.49 million of taxable income filing jointly. It is a real case and a narrow one.

These tables also give BOXX no credit for deferral, which is real and which compounds. The calculator adds it back.

The state map

A single state rate hides most of what is going on.

Eight states levy no individual income tax at all (Alaska, Florida, Nevada, New Hampshire, South Dakota, Tennessee, Texas and Wyoming). There, BOXX faces the full federal edge with no offset. Missouri joins them in effect: it has exempted capital gains entirely for tax years from 2025, and SGOV’s Treasury interest was already exempt everywhere by federal preemption. BOXX ends up marginally ahead there rather than level, because the small non-government slice of SGOV’s distribution is still Missouri-taxable while BOXX’s gain is not. On the modeled inputs that is worth about a basis point.

Washington is the exception among the no-income-tax states. It taxes no wages and no interest, so an SGOV holder pays nothing. It does levy an excise on long-term capital gains at 7%, plus an additional 2.90% on gains above $1 million, after a standard deduction of $278,000 for tax year 2025. The 2026 figure is not published until the end of October. Most Washington investors are below that deduction and face a 0% marginal rate, which puts them alongside Texas. An investor realizing large gains in the same year, from a business sale or a concentrated position, faces 7% on BOXX and nothing on SGOV.

Washington also enacted a separate 9.90% income tax in March 2026, effective for 2028 and first payable in 2029, which is a different tax with confusingly similar parameters. It reaches households with more than $1 million of Washington base income, and that deduction is per household rather than doubled for couples. Below that threshold nothing in the comparison above changes. Above it, interest on federal obligations keeps an express deduction, so SGOV stays exempt, while long-term gains are pulled out of the base and added back for taxpayers who owe the capital gains excise, with a credit for the excise paid that is capped at the income tax otherwise due. The practical effect is to widen the existing Washington asymmetry at the very top. The Department of Revenue has published no implementation guidance, and a constitutional challenge is pending with no ruling on the merits.⁶

Seven states give long-term gains a genuine preference that reaches fund investors: Arizona, Arkansas, Hawaii, Montana, North Dakota, South Carolina and Wisconsin. Several other states are commonly listed as having preferences that in fact apply only to business sales or in-state property, and those do not help an ETF holder.

Two states widen the gap. Massachusetts adds a 4% surtax above $1,107,750 for 2026, taking long-term gains to 9%. Minnesota adds a 1% tax on net investment income above $1 million, taking it to 10.85%.

One mechanical detail worth knowing if you hold Treasury funds: California, Connecticut and New York each require a fund to hold at least 50% of assets in federal obligations at every quarter-end before shareholders may claim the state exemption. SGOV meets that test. Some broader government funds do not.

Section 1258, and the three-way disagreement

Congress enacted §1258 in 1993 to stop conversion transactions: arrangements that produce capital gain when substantially all of the return is really the time value of money. The test has two parts. First, substantially all of the expected return must be attributable to time value. Second, the transaction must fall into one of four categories, including a contemporaneous purchase-and-forward-sale, an applicable straddle, a transaction “marketed or sold” as producing capital gains from a time-value return, and anything Treasury specifies by regulation.

Everyone agrees the first part is met. The disagreement is entirely about the second part, and about which taxpayer is even in the transaction. Three published positions, and they are genuinely different.

Steven Rosenthal: the shareholder is in a conversion transaction

Rosenthal, of the Tax Policy Center, notes that he helped draft §1258 as a Joint Committee on Taxation staffer in 1993. He argues that “substantially all of the shareholders’ expected return is attributable to the time value of their net investment,” and that the second prong is satisfied through the marketing category, with Treasury’s regulatory catch-all available as a backstop.⁸

His proposed remedy is rulemaking rather than enforcement: “even if the marketing of BOXX did not establish investment in the fund as a conversion transaction, regulators could still proceed under the catchall fourth provision.” He does not allege bad faith by the sponsor, and he grants the appeal of the trade. If he is right, gain on sale becomes ordinary income and the deferral benefit survives.

Daniel Hemel: the problem is at the fund, and it is worse

Hemel, of NYU, explicitly declines the shareholder theory and aims at the fund, invoking §1092’s straddle rules alongside §1258. His published abstract puts it plainly: “this article concludes that the IRS would likely succeed in recharacterizing gains from BOXX’s trades as ordinary income at the fund level. Whereas shareholder-level recharacterization would leave BOXX’s deferral benefits in place, fund-level recharacterization would erase both the deferral and rate-arbitrage benefits that BOXX advertises.”⁹

That distinction deserves emphasis because it changes the stakes. Rosenthal’s theory, if it prevailed, leaves a BOXX holder roughly where a T-bill holder already is. Hemel’s leaves them worse off, which is why he titled the piece around a trap.

Mark Leeds: the shareholder is fine, and the fund has no gain

Leeds, of Mayer Brown, published the longest technical treatment. He concedes the first prong at both levels, calling it “very difficult to argue” otherwise. He then rejects three of the four second-prong categories for a shareholder: absent a forward contract to sell the shares, the first category fails; “the holding of the BOXX stock should not be considered a straddle under section 1092”; and no Treasury regulation designates anything, which renders the fourth category moot.¹⁰

On the marketing category he is more careful, calling it “more complicated because it seeks to discern what the investor was told,” noting press interviews discussing tax efficiency, and concluding that “while some uncertainty exists, the test in section 1258(c)(2)(C) should not be considered satisfied based on what has occurred to date.” Both hedges are his.

At the fund level he concedes more than the sponsor might like, allowing that the fund’s single-stock option positions probably do create an applicable straddle. Then comes his structural answer: §1258(a) recharacterizes gain that is recognized, and by using §852(b)(6) the fund disposes of appreciated positions in kind without recognizing any. “Even though BOXX itself may have engaged in a conversion transaction, there is no gain at the BOXX level for section 1258(a) to recharacterize as ordinary income.”

Leeds also spends pages arguing against himself, listing the doctrines the IRS could deploy. His conclusion is qualified rather than reassuring: “While the strategy has some tax uncertainty, it appears strong enough to be offered to investors with adequate disclosure of the risks.”

One disputed sentence, and one empty shelf

Rosenthal and Leeds disagree about a verifiable fact. Rosenthal says the prospectus describes BOXX as an opportunity to earn interest-like returns with capital-gains taxation. Leeds says the offering materials “do not market the BOXX shares as producing capital gains.” The sentence at issue reads: “Based on the advice of its accountants, the Fund expects that distributions related to the Fund’s SPX positions, if any, will be characterized by the Fund as capital gains with these preferential terms.”¹ It is a statement about fund distributions, hedged twice, and it sits near a risk factor naming both theories. Read it and decide; both readings are defensible.

The deeper reason nothing is settled is that in more than thirty years Treasury has issued exactly one regulation under §1258, a computational netting rule that designates no transaction as anything. Leeds calls the catch-all moot because it has never been used. Rosenthal calls it the fix because it is available. They are describing the same empty shelf.

What Alpha Architect says

The fund does not claim the question is settled. Its statement of additional information says it “has received guidance from tax counsel regarding section 1258 of the Code… but such guidance is not necessarily persuasive or binding on the IRS. Notwithstanding the guidance, the IRS might seek to recharacterize the tax consequences at the Fund level, the shareholder level, or both. A successful challenge could have a negative effect on the Fund, its shareholders, or both.”¹¹

Wes Gray, Alpha Architect’s CEO, has been consistent in public about not discussing BOXX taxation. On the Bogleheads on Investing podcast in June 2024 he told Rick Ferri: “One of our core beliefs is transparency. However, when it comes to BOXX, I always tell people it’s like Fight Club. The first rule of BOXX Club is we cannot talk about taxes. And the reason for that is to protect shareholders… There’s this rule. It’s called Section 1258.”¹²

That is a coherent compliance posture aimed at the marketing category, and it is also an implicit acknowledgment that the category is live. The current investment-case deck backs it up: its three selling points are access to the box market, professional management and reasonable costs, with no tax claim anywhere, alongside a statement that neither Alpha Architect nor its affiliates provide tax advice.⁴

Gray has made one substantive public argument that deserves airing on its own terms, about §852(b)(6) as policy: mutual fund shareholders can inherit somebody else’s tax bill when other investors redeem, and in-kind redemption removes that externality. In the same conversation he volunteered that other preferences, naming carried interest and private placement life insurance, should be closed. He also described hedging against legislative change by banking losses inside the fund.¹²

Elm’s ledger

Victor Haghani and James White of Elm Wealth published the most careful economic critique in February 2024. It is not a hostile document. They thank Wes Gray and Alpha Architect’s trading director for discussing the fund with them before publication, and they open by noting “we are not tax experts,” treating the tax question as one risk among several rather than adjudicating it.¹³

Their method is to start with the tax prize and subtract everything else. At the roughly 5% short rates of early 2024, the seventeen-point rate difference was worth 0.85% a year, plus another 0.07% for five years of deferral. Then the deductions: 0.21% for the fee gap against a T-bill ETF, 0.5% for the return options counterparties demand to take the other side, 0.5% for state taxes at a roughly 10% state rate, 0.2% to 0.5% for clearinghouse credit risk, and 0.25% for a 30% chance of needing the cash inside a year. Their conclusion: “Adding the cost of credit risk to all the other costs already noted gives us total costs greater than the 0.85% potential tax benefit.”

The counterparty line is the most interesting argument in the piece. They accept the historical evidence that box rates run above Treasury rates, then ask what happens next: “it is worthwhile to ask yourself how you would price those options if you were being asked to be the counterparty to BOXX’s trades?” A dealer selling boxes must post collateral, buys a T-bill to do it, and therefore needs to price the box below that bill to earn a spread. Alpha Architect reads the same historical premium as a durable edge the fund harvests. Nobody has published a measurement covering the period since, and BOXX has grown roughly thirteenfold in the meantime. That is an open empirical question.

Two of Elm’s inputs have moved

Their fee line used BOXX’s then-stated long-term expense ratio of 0.395%, noting a waiver at 0.195% running to January 31, 2025. The current prospectus carries a 0.2449% gross management fee with a waiver to 0.1949% running to at least February 1, 2027.¹ The specific concern that the fund would revert toward 0.395% no longer matches the fee schedule.

Their liquidity concerns were framed around a fund with about $1 billion in assets. BOXX now holds $13.77 billion with a one-basis-point median secondary spread. Underlying FLEX liquidity and authorized-participant risk are still disclosed, but a small thinly traded ETF is no longer an accurate description.

The rest of their ledger is untouched by those two updates. State taxes, the counterparty spread, clearinghouse credit risk and the early-sale problem are all unchanged, and their framework remains the right way to think about it. Their PDF carries a 2026 upload path, but the document is the February 27, 2024 original; they have published no update.

Their conclusion is also more measured than it is usually reported. They wrote that they were “not at present” planning to use BOXX for clients, “especially those who are subject to high rates of state taxation,” and named an explicit exception: an investor whose capital loss carryforwards are so large they will never be fully used. Their deepest worry is about ETF taxation generally: they argue in-kind redemption is fairer to long-term investors than the mutual fund treatment it replaced, and they fear a fund large enough to attract attention could precipitate a rule change that costs everyone. The piece ends on the line “We truly hope our worries are misplaced.” Gray argues the same policy point from the other direction.

Ranking the risks

• Tax-law risk is the largest and the least quantifiable. Note the mechanic buried in the prospectus: if the fund lost RIC status it might pay a deficiency dividend, and that dividend is paid to “the then current shareholders,” not to whoever held it when the gain arose. The same disclosure warns that a recharacterization could leave shareholders “having underreported income or gains to the IRS for the applicable years.”¹
• State tax is a computable drag rather than a risk. You can work it out before you invest, and the tables above do.
• Clearinghouse exposure is real and modest. Options are guaranteed for settlement by the OCC, which removes bilateral counterparty risk. The prospectus still warns that “in the unlikely event that the OCC becomes insolvent or is otherwise unable to meet its settlement obligations, the Fund could suffer significant losses.” A Treasury bill is an obligation of the United States; this is not. Elm’s 0.2% to 0.5% charge for that difference is an estimate rather than an observable price, and reasonable people put it lower.
• FLEX options can be hard to value. The prospectus notes they “may be less liquid than other securities” and that potentially a large portion of the portfolio may be valued on bases other than market quotations.
• Selling inside a year removes most of the point. A short-term gain is taxed at ordinary rates, which erases the character benefit and leaves you paying the higher expense ratio for deferral alone. Calling BOXX an emergency fund glosses over that.
• Low rates shrink the prize proportionally. A seventeen-point rate spread on a 5% yield is worth 85 basis points; on a 1% yield it is worth 17. The prospectus goes further and warns the fund “may have a negative yield (i.e., it may lose money on an operating basis)” if short rates fall far enough.

Cheaper competitors have arrived

Two rival box-spread ETFs registered in August 2026. GraniteShares launched LBOX with a 0.1349% net expense ratio, six basis points under BOXX, with its waiver running through December 2027. Tidal registered XCSH, copying BOXX’s stated objective word for word, at roughly 0.205% net.¹⁴

Both are days old with no assets and no track record, so treat them as evidence that fee competition has started rather than as alternatives with a history. They also carry the same unresolved tax question, since the structure is what creates it.

Who BOXX fits

Work down this list. The more answers on the left, the stronger the case.

Strong candidate. Taxable money that will sit for several years, a wide gap between ordinary and long-term rates, and little or no state exposure. A high earner in Florida or Texas is the clean case, as is a Washington investor whose other realized gains stay under the deduction. Large unused capital loss carryforwards make the case considerably stronger, because they can absorb capital gain in a way they cannot absorb unlimited interest income. This is the point Rick Ferri made in 2023 and Elm conceded in 2024. Every figure in this bucket assumes the intended capital-gain treatment survives. Read the modeled edge as payment for accepting complexity and an unresolved statute.

Close call. A modeled edge in the range of ten to twenty-five basis points. At that size, simplicity, direct Treasury backing and a settled tax treatment are worth something, and splitting between the two funds is defensible.

SGOV. Money that may be needed inside a year, a narrow rate spread, a tax-advantaged account, or no interest in tracking a tax argument. Treasury bills remain the cleaner instrument for a genuine emergency reserve, and the cash guide walks through the layers.

What I would do

Make Treasury bills or SGOV the default and treat BOXX as an optimization layer you can justify by name. If you cannot articulate which specific feature of your tax situation BOXX exploits, the answer is SGOV. Complexity should have to earn its place.

For disclosure: I hold a modest BOXX position, roughly $20,000, alongside a considerably larger SGOV position that holds most of my cash. That shape is the recommendation. It captures some of the tax upside without making a liquidity allocation depend on the most favorable reading of an unresolved statute.

How Summitward helps

Cash allocation is a tax question before it is a yield question, and the right answer moves with your bracket and your state.

Frequently asked questions

Is BOXX safe?

The strategy holds fully hedged option positions cleared through the OCC rather than obligations of the U.S. Treasury. Held to expiration the payoff is fixed, so directional market risk is close to nil, though the prospectus notes a box sold early can lose money as rates move. What you are accepting instead is clearinghouse exposure, valuation and liquidity risk on customized options, and tax uncertainty. These are different in kind from the risks of a T-bill fund rather than larger versions of them, and in normal conditions the first two are small. The tax risk is not small, and none of the three is easy to put a number on.

How is BOXX taxed?

The fund aims to distribute nothing, and in most years it has, so there is usually nothing to tax until you sell. On sale, appreciation held more than a year is intended to be long-term capital gain. Whether that character holds is the disputed question. In one year of four the fund did distribute, and that distribution was taxable to whoever held shares on the record date.

Does BOXX make sense in an IRA?

No. The advantage is entirely about the character and timing of taxable income, and neither exists inside a tax-advantaged account. You would be paying a higher expense ratio and accepting option-structure risk for nothing.

Can I use it as an emergency fund?

Day to day the shares trade with a one basis point median spread, so selling is easy in ordinary conditions. Two things complicate it. The prospectus warns that FLEX positions can become difficult to value and exit under stress, which is when an emergency reserve actually gets used. And selling inside a year makes the gain short-term and taxed at ordinary rates, which removes the reason for holding it. Elm made the second point sharply: needing instant access is at odds with a strategy that requires a holding period.

What happens if the IRS challenges it and wins?

It depends which theory prevails. Shareholder-level recharacterization would make your gain ordinary, leaving you roughly where a T-bill holder already sits and preserving deferral. Fund-level recharacterization would remove both benefits. The prospectus also raises a scenario where the fund pays a deficiency dividend to then-current shareholders, which means the person holding when a challenge succeeds may bear consequences from an earlier period.

Is the tax benefit bigger in California or in Texas?

Texas, by a wide margin. California taxes capital gains at ordinary rates, so a California investor gives back most or all of the federal edge unless their ordinary-to-long-term spread is at the very top. Run your own break-even in the calculator above rather than relying on a rule of thumb.

Why does BOXX not publish an SEC yield?

Because there is almost no accounting income to report. Net investment income per share has been negative every fiscal year, since expenses exceed the small amount of recognized income. The fund publishes an average yield to option expiration instead, which is forward-looking and excludes fees, so it is not comparable to an SEC yield.

Should I worry that BOXX trailed its benchmark?

It is worth knowing, and the detail cuts both ways. In the audited annual report through September 30, 2025, BOXX returned 4.91% annualized since inception against 4.99% for the Solactive 1-3 Month US T-Bill Index, eight basis points behind. Over the one-year period in that same table it was ahead, 4.51% against 4.48%. And the eight-point gap is smaller than the 19.5 basis points the fund charges, so gross of fees the strategy ran ahead of the index and the fee is what turned that into a small deficit. The fund page now shows 4.72% since inception as of July 31, 2026, but publishes no benchmark alongside it, so there is no current gap to quote. What all this means is that the case for choosing BOXX over Treasury bills rests on the tax treatment, since the pre-tax economics land close enough to a wash.

Key takeaways

• Three questions, three answers. Box spreads work, BOXX’s after-cost case depends entirely on your situation, and the tax character is unresolved. Confidence about one tells you nothing about the others.
• The case for choosing it over T-bills is an after-tax case. Through September 2025 the audited record showed the fund eight basis points a year behind its own T-bill benchmark since inception, while charging 19.5, so the strategy itself ran ahead of the index before fees and the fee closed the gap.
• State residence can flip the answer. Treasury interest is exempt from state tax under 31 U.S.C. §3124(a) and capital gains are not, so the break-even state rate runs from roughly 5% at a 22%/15% federal situation to roughly 16% at 37%/20% plus NIIT on the modeled assumptions.
• The experts disagree in three directions. Rosenthal targets the shareholder, Hemel targets the fund and reaches a worse outcome for investors, and Leeds argues §852(b)(6) leaves no recognized gain for §1258 to reach. Treasury has issued one regulation under the statute in thirty years and it designates nothing.
• Elm’s framework holds and two of its inputs have moved. The fee gap has narrowed and the fund is no longer small or thinly traded, while the state-tax, counterparty and early-sale arguments are unchanged.
• Loss carryforwards are the clearest case. An investor with large unusable capital losses can absorb BOXX gains in a way they cannot absorb interest income, which is the one exception both Rick Ferri and Elm identified independently.

Related guides

• Alpha Architect ETFs: the factor funds from the same firm, and how they differ from everything else under the brand
• Box Spread Loans covers the same instrument from the borrowing side, with the four-leg algebra and why the payoff is exact.
• Where to Park Your Cash places SGOV, T-bills, money market funds and munis into reserve layers.
• SGOV vs USFR compares the two main Treasury cash ETFs with an after-tax calculator.
• Four Things You Can Do With a Capital Gain develops the deferral math and the §852(b)(6) in-kind mechanism.
• Tax-Loss Harvesting covers how carryforwards accumulate and what they can offset.
• Options Risk and Reward covers §1256 treatment for directional options trades.

Sources and method

1. Alpha Architect 1-3 Month Box ETF, Prospectus and Statement of Additional Information, dated February 1, 2026, filed with the SEC by EA Series Trust. Source for the zero-coupon bond comparison, the sub-90-day weighted average maturity, the European-style option policy, the §852(b)(6) in-kind language, the §1258 risk factor, the OCC insolvency disclosure, the FLEX options liquidity and valuation risks, the low short-term interest rate warning, the fee waiver running to at least February 1, 2027, the deficiency dividend and underreporting language, the “lack of clear guidance” tax risk factor, the SPX §1256 sentence, the reservation of the right to use other underliers, and the financial highlights showing negative net investment income per share. SEC EDGAR
2. Alpha Architect, BOXX fund page, figures as of August 17–18, 2026: $13,769.04 million in net assets, 0.2449% gross and 0.1949% net expense ratio, 30-day median bid/ask spread of 0.01, average 34 days to option expiration, average yield to option expiration of 4.00% (a forward figure the fund states excludes fees, expenses and commissions), inception December 27, 2022, listed on Cboe BZX. Also the source for the European-exercise ticker convention and the current SPY and QQQ FLEX holdings. Alpha Architect
3. EA Series Trust, Annual report and audited schedule of investments, fiscal year ended September 30, 2025 (Form N-CSR). Purchased options were 101.4% of net assets, consisting of SPDR S&P 500 ETF Trust and Invesco QQQ Trust Series 1 FLEX options. Average annual total returns: BOXX at NAV 4.51% for one year and 4.91% since inception on December 27, 2022, against the Solactive 1-3 Month US T-Bill Index at 4.48% and 4.99%. Total net assets $8,079,665,315. SEC EDGAR
4. Alpha Architect, Investment Case: Cboe:BOXX, data as of June 30, 2026. The three stated selling points are access to box spreads, professional management and reasonable costs, with no tax-benefit claim in the document, alongside the statement that neither Alpha Architect nor its affiliates provide tax advice. Also the reference to Billingsley and Chance, “Options Market Efficiency and the Box Spread Strategy,” Financial Review, 1985. Alpha Architect
5. Internal Revenue Service, Revenue Procedure 2025-32. Section 3.03 sets the 2026 long-term capital gain breakpoints: $98,900 and $613,700 for married filing jointly, $49,450 and $545,500 for single filers, $66,200 and $579,600 for heads of household. Section 2 confirms the seven ordinary rates topping at 37% remain in effect following Public Law 119-21, enacted July 4, 2025. The 3.8% net investment income tax under §1411 applies above $250,000 of modified AGI filing jointly and $200,000 single, reaches interest and capital gains alike, and the IRS states those thresholds are not indexed for inflation. IRS (PDF)
6. 31 U.S.C. §3124. Subsection (a): “Stocks and obligations of the United States Government are exempt from taxation by a State or political subdivision of a State. The exemption applies to each form of taxation that would require the obligation, the interest on the obligation, or both, to be considered in computing a tax,” with exceptions only for a nondiscriminatory corporate franchise tax and estate or inheritance taxes. Subsection (b) assigns “the tax treatment of gain and loss from the disposition of those obligations” to the Internal Revenue Code. California and Connecticut each define the exempt dividend as one other than a capital gain dividend, and both states, along with New York, require a fund to hold at least 50% of assets in federal obligations at each quarter-end. Washington’s capital gains excise is at RCW 82.87.040 as amended by ESSB 5813 (Ch. 421, Laws of 2025), imposing 7% plus an additional 2.90% on gains above $1 million, with the standard deduction at RCW 82.87.060 indexed under RCW 82.87.150 and published by October 31 each year. Washington’s separate individual income tax is ESSB 6346 (Ch. 238, Laws of 2026): 9.90% from January 1, 2028, a $1,000,000 household standard deduction at section 314, the federal obligations deduction at section 306, the long-term capital gains modification at section 302, and the nonrefundable credit for capital gains excise tax paid at section 205. Missouri’s capital gains exemption is at Mo. Rev. Stat. §143.121.3(14)(a) for tax years from 2025. Cornell LII
7. iShares, 0-3 Month Treasury Bond ETF (SGOV), figures as of August 17–18, 2026: $103.59 billion in net assets, 0.09% expense ratio with no waiver, 30-day SEC yield 3.60%, average yield to maturity 3.71%, effective duration 0.11 years. U.S. government source income was 95.14% for tax year 2025 per BlackRock’s annual state tax reporting. The SEC yield is net of expenses and backward-looking over a completed 30-day window; average yield to maturity is gross of fees, which is why the two families of figure should not be subtracted from one another. iShares
8. Steven Rosenthal, Tax Gimmick in a BOXX, Tax Policy Center TaxVox, and BOXX’s Tax Gimmick Violates Congress’ Rules On Conversion Transactions, Forbes, March 4, 2024. Source for his statement that he helped draft §1258 as a Joint Committee on Taxation staffer in 1993, his argument that the shareholder’s expected return is time value, his reliance on the marketed-or-sold category, and his proposal that Treasury proceed under the regulatory catch-all. Tax Policy Center
9. Daniel J. Hemel, The Tax Trap Inside the BOXX, 182 Tax Notes Federal 1973 (March 11, 2024). Quoted from the published abstract; the article body is behind a paywall and is not quoted here. Hemel concludes that the IRS would likely succeed in recharacterizing gains at the fund level rather than the shareholder level, and that fund-level recharacterization would erase both the deferral and the rate arbitrage. Tax Notes
10. Mark Leeds, The BOXX Anomaly: ETF Tax Enhances Inverted Yield Curve Strategy, 185 Tax Notes Federal 1541 (November 25, 2024), authorized reprint hosted by Mayer Brown. Source for his concession on the time-value prong, his rejection of the forward-sale, straddle and Treasury-designation categories at the shareholder level, his treatment of the marketing category as unsatisfied “based on what has occurred to date,” his §852(b)(6) argument that there is no recognized gain for §1258(a) to reach, and his conclusion that the strategy “appears strong enough to be offered to investors with adequate disclosure of the risks.” The related NYSSCPA presentation with Daniel Hemel, December 2024, explicitly flags where the two authors differ. Treasury’s only regulation under §1258 is 26 C.F.R. §1.1258-1, a netting rule. Mayer Brown (PDF)
11. EA Series Trust, Statement of Additional Information, February 1, 2026, Conversion Transactions section: the fund “has received guidance from tax counsel regarding section 1258 of the Code and its applicability to the current investment strategy of the Fund, but such guidance is not necessarily persuasive or binding on the IRS. Notwithstanding the guidance, the IRS might seek to recharacterize the tax consequences at the Fund level, the shareholder level, or both. A successful challenge could have a negative effect on the Fund, its shareholders, or both.” SEC EDGAR
12. Wesley Gray, interviewed by Rick Ferri, Bogleheads on Investing, episode 70, June 3, 2024. Source for the “first rule of BOXX Club” passage and its reference to §1258, the §852(b)(6) policy argument about mutual fund shareholders inheriting other investors’ tax bills, his remark that carried interest and private placement life insurance should be closed, and his description of banking losses inside the fund as a hedge against legislative change. Ferri notes on air that some academics believe the gains should be taxed as ordinary income. Bogle Center
13. Victor Haghani and James White, Elm Wealth, Thinking Outside the BOXX, February 27, 2024. Source for the full deduction ledger: 0.85% gross federal benefit at roughly 5% short rates, 0.07% for five years of deferral, 0.21% for the fee gap against BIL using BOXX’s then-stated 0.395% long-term expense ratio, 0.5% for counterparty required margin, 0.5% for state taxes at a roughly 10% state rate, 0.2% to 0.5% for clearinghouse credit risk, and 0.25% for a 30% chance of an early sale. Also the acknowledgment that Wes Gray and Larry Lempert discussed the fund with them, the statement “we are not tax experts,” the loss-carryforward exception, the conditional conclusion that they are “not at present” using BOXX for clients, and the closing hope that their worries are misplaced. The PDF served from a 2026 upload path is the same February 2024 document. Elm Wealth
14. GraniteShares Short Term Box ETF (LBOX), Form 497K, prospectus dated August 14, 2026: 0.15% management fee less a 0.0151% waiver for a 0.1349% net expense ratio, waiver running through December 31, 2027, listed on Cboe BZX. XFUNDS 1-3 Month BOX ETF (XCSH), Tidal Trust II, prospectus dated August 10, 2026, listed on Nasdaq; the fee table shows 0.205% net while a footnote states 0.18%, and that inconsistency is unresolved in the filing. Neither fund had reported assets when this was written. SEC EDGAR
15. Jules van Binsbergen, William Diamond and Peter Van Tassel, Options for Calculating Risk-Free Rates, Federal Reserve Bank of New York Liberty Street Economics, October 2, 2023: “from January 1996 to April 2023, the box rate is 35 basis points above the Treasury rate on average, implying a 35 basis point convenience yield,” peaking near 130 basis points in October 2008 and running between 20 and 40 basis points in recent years, with a nearly flat term structure out to three years. The related paper by van Binsbergen, Diamond and Grotteria, Journal of Financial Economics 2022, measures roughly 40 basis points over an earlier sample and finds the convenience yield larger below three months maturity. The two share authors and methodology, so they are one research program at two vintages rather than independent confirmation. Liberty Street Economics
16. Rick Ferri, post to the Bogleheads forum thread “AlphaArchitect launches BOXX: 1-3 Month Box Spread ETF,” February 2, 2023, quoted from a screenshot of the thread. His caveat is included in the original: “I’m not a CPA. Consult your tax adviser.”
17. Method. The comparison arithmetic lives in web/src/lib/boxx-after-tax.ts and is unit tested against every cell of the tables above. Both funds are given the same 4.00% gross return so the comparison isolates fees and taxes; expense ratios are 0.1949% and 0.09%; SGOV is treated as passing through 95.14% of its distribution free of state tax and paying ordinary federal rates on all of it; BOXX is treated as producing long-term capital gain taxed once on sale. The 3.8% net investment income tax is applied to both funds when enabled. The break-even state rate is solved in closed form from those inputs. Nothing here models trading costs, tracking error, the possibility of an unwanted distribution, bracket boundaries crossed by the return itself, states with a preferential capital gains rate, or the tax-law risk that is the subject of half this guide. Every table figure is a modeled output under stated assumptions rather than a yield anyone will receive. Nothing here is a forecast, and nothing here is tax, legal or investment advice.