Box Spread Loans: What They Save Depends on the Rate You're Escaping

Somebody on a forum mentions they are borrowing at 4.3% against their brokerage account by selling four option contracts at once. You check your own broker’s margin schedule and find 11.825%. The gap is seven and a half percentage points, and the machinery producing it is put-call parity, which has been in textbooks since the 1970s.

The trade works as advertised. Cboe reports that daily notional lending volume in SPX box spreads has more than tripled in three years, hitting a record $1.7 billion in 2026, with RIAs and family offices adopting them as a cheap alternative to margin loans.¹ The instrument works, it clears through OCC, and the rate is genuinely close to what large institutions pay to fund themselves.

The question almost nobody asks is what you would have paid otherwise. Every article on this topic benchmarks the box against a double-digit retail margin rate. Against a Pledged Asset Line the gap is a third of that. Against Interactive Brokers’ own margin rate, on the same day, a one-year box saves about a quarter of a percentage point. Which rate you are escaping decides whether this saves you several percentage points or roughly $700 a year.

What a short box spread is

A box combines four European options on the same index with the same expiration, at two strikes. To borrow, you take the short side:

• Sell the lower-strike call and buy the lower-strike put
• Buy the higher-strike call and sell the higher-strike put

That is a synthetic short position at $K_1$ paired with a synthetic long at $K_2$. Add the four payoffs at a settlement level $S$ and the index cancels:

The result holds for every value of $S$, including values exactly at a strike. You receive a credit today and owe $(K_2-K_1)\times 100$ per contract at expiration. On the 1,000-point width that carries 92% of SPX box volume, that is $100,000 owed per contract.²

Schwab published a worked example in November 2025. On October 16, 2025, a short box using the SPX 6,600 and 6,700 strikes expiring March 20, 2026 collected a $9,830 credit against a $10,000 settlement obligation. Whether SPX finished at 6,500, 6,650 or 6,800, the amount owed was $10,000. The $170 difference was the financing cost, which Schwab annualized at about 4.15%.³

Why the rate sits close to wholesale funding

Because the equity exposure cancels, a box price is a discount factor wearing a costume. Two option strikes, one expiration, and what is left after the algebra is the market’s price of time. That is why academics use boxes as a measuring instrument rather than a trade. Van Binsbergen, Diamond and Grotteria inferred risk-free rates from put-call parity on index options precisely to get a rate uncontaminated by the convenience yield investors pay for holding Treasuries, and measured that convenience yield at roughly 40 basis points on average, larger inside three months, quadrupling during the financial crisis.⁴

That finding cuts both ways for a borrower. It explains why box rates can beat retail lending: you are transacting in a market where institutional dealers set the price. It also sets a floor. A box borrower pays the convenience-yield-free rate by construction, so a box will structurally cost more than a Treasury of the same maturity, and the gap widens in a crisis, which is exactly when you would want to roll one.

The observed numbers match. Across 2025, Cboe measured SPX box lending rates averaging about 26 basis points over 3-month Treasuries, ranging 6 to 44 basis points in the first half.² In August 2026 Cboe’s derivatives market intelligence desk put the premium at 69 basis points, a multi-year high against a historical average near 32, attributing the widening to constrained dealer balance sheets meeting rising borrowing demand.¹ Read that figure carefully: the chart behind it plots a three-month box yield against three-month SOFR, so it is a premium at one point on the curve rather than a spread you can bolt onto an overnight rate.

Boxes have a term structure like anything else, and Cboe publishes a separate box rate index for each listed SPX expiration, which removes the need to estimate at all. On August 19, 2026 the November 2026 series stood at 4.16%, the September 2027 series at 4.45%, and the December 2031 series at 4.64%.¹⁵ Those are the numbers used below. Look up the series matching the expiration you are actually pricing rather than adding a spread to SOFR, and treat any published figure as an indication until you see a fill.

The comparison that decides it

Here is where the published material and the arithmetic part company. Cboe’s own market data page compares a 4.19% box against a median broker margin rate of 10.950%, a spread of nearly seven points. The same page also discloses that broker rates for a $100,000 loan ranged from 5.83% to 11.075%.² The median is the headline; the low end of that range is the number a rate-shopping borrower would face.

All rates below are as published in mid-August 2026, on the same borrowing of $250,000 for one year.

Interactive Brokers is the broker where box spreads are easiest to execute and where most of the forum discussion happens, and it is also the broker where the box saves the least. Note that IBKR blends its tiers rather than charging one rate on the whole balance, so a $250,000 loan costs 4.83% and not the 4.63% its second tier advertises. Against that, a 38 basis point edge on $250,000 is $950 a year. That is what the options approval, the four-leg execution, the fixed maturity and the tax complexity are buying.

The managed-service arithmetic makes the point sharper. Firms that execute and roll boxes on your behalf charge on the borrowed amount. Add half a point of fee to a 4.45% box and you are at 4.95%, which is 12 basis points above IBKR’s own blended rate for a loan of that size, with no options approval required and no execution risk.

None of that argues against the instrument. It argues for running the comparison against the rate you can actually get rather than against the worst rate in the market.

Three ways to quote the same rate

Before you can run that comparison you have to agree on what “the rate” means, and the published material does not. Three conventions are in live use and they disagree by 20 to 45 basis points on the same trade.

• Simple, ACT/365, over proceeds. Financing cost divided by cash raised, times 365/days. OCC’s 2025 box-spread paper computes its example this way, getting 4.16% on a $96,000 debit against a $100,000 payoff one year out.⁶
• Bank discount, ACT/360, over face. The Treasury-bill convention. Divides by the amount owed rather than the amount received, on a 360-day year, so it always reads lowest.
• Compounded, ACT/360. The internal rate of return from proceeds to obligation. This is the one that compares like for like against a margin loan, an SBLOC or a HELOC, all of which quote a nominal rate accruing on an outstanding balance over a 360-day year.

OCC’s 2020 primer manages to use two of them four pages apart. It computes the box rate as $[(1000-999.40)/999.40](365/46) = 0.48\%$, dividing by price on a 365-day year, and then tells readers that “the effective rate on a box represents a ‘discount yield’ similar to a quoted T-bill rate,” illustrating with a calculation that divides by face on a 360-day year. The same document also transposes its own bid and offer labels: the quote is given as 999.35 bid and 999.40 offered, then the rate section reads “Bid at 999.40” and “Offered at 999.35.”⁷ The economics in the surrounding text are right; the labels are reversed.

The gap between conventions grows with the term, because simple annualization ignores compounding. A box bought at 800 against a 1,000 payoff five years out reads 5.00% simple and 4.56% compounded, a difference of 44 basis points. That is the tenor used to pitch box spreads as mortgage substitutes, and it is the tenor where the choice of convention moves the answer most.

Put a quote in and see all four, alongside whatever your own alternatives cost you.

The collateral is what moves

A correctly built SPX box has a terminal value known to the penny. The portfolio backing it does not. Cboe’s advisor guidance, written by its own director of derivatives sales rather than by a vendor, names this directly: “The biggest risk is the asset-liability mismatch. The amount owed on a box spread is set with a fixed term, but the securities supporting the borrowing are marked to market each day. If asset values fall, the client may be required to post additional collateral or face forced liquidation of their holdings.”⁸

Three specific mechanisms deserve attention, and only the first is widely discussed.

A collateral stress test. Broker margin treatment of a short box is account-specific and differs sharply between Reg T and portfolio margin, so the figure below is a generic leverage bound rather than your liquidation threshold. With collateral $V$, a loan $L$, and a maintenance requirement that equity stay above a fraction $m$ of value, the market decline you can absorb is $1 - (L/V)/(1-m)$. Borrow half your portfolio against a 30% equity minimum and a 28.6% decline puts you at the line. Broker house requirements exceed the regulatory minimum and can be raised without advance notice, so treat that as the optimistic bound.

The obligation does not appear as a margin balance. A margin loan shows up as a debit on your statement, accruing interest you can see. A short box shows up as four option positions. Net liquidation value barely moves at execution, because the cash raised and the liability assumed are close to equal. Nothing on the screen says “you have borrowed $250,000,” which makes it unusually easy to borrow more than you meant to and to forget how much is outstanding.

Bad marks can trigger real calls. Four thinly quoted legs get marked individually, and the sum can wander well away from the box’s economic value. Schwab warns that box spreads “can sometimes cause an account to show losses, restricting other opportunities for traders due to a reduced net liquidation value,” and that liquidity issues make marking all four legs at the desired prices difficult for brokerages and exchanges.³ Margin systems act on current marks, so a bad mark is a live problem even when the terminal payoff is not in doubt.

There is also a capital-efficiency question that gets underplayed. Under Reg T, a short box consumes buying power in a way that has little to do with its risk. The Cboe-hosted piece promoting long-dated boxes concedes that a $1.25 million box “requires at least a $2.5MM portfolio as collateral” under Reg T.⁹ Cboe’s own advisor piece adds that in Reg T accounts short boxes “can be capital intensive and may limit flexibility unless handled by experienced traders,” and that risk-based portfolio margin “often aligns better with the defined nature of a box spread.”⁸ Portfolio margin generally requires a six-figure account minimum, which quietly excludes a large share of the people the 4% headline attracts.

Why the payoff is exact, and what can still break it

SPX options are European, so they cannot be exercised before expiration, and they are cash-settled, so nothing is ever delivered. OCC states the reason plainly: European-style options “ensure that the box spread cannot be exercised early which would result in the cancellation of the effective loan before the term date.”⁷

There is a second layer of protection that rarely gets mentioned, provided every leg is the same product. Four legs of standard SPX settle against one exercise-settlement value, the SET print, built from the opening prices of the 500 component stocks on expiration morning.¹⁰ SET is a level nobody can trade at, which is a genuine basis risk if you hold a single SPX option into expiration. In a box that risk cancels exactly, because every leg references the identical print. The algebra above evaluates to $-(K_2-K_1)$ at every value of $S$, so pin risk and exercise-by-exception thresholds have nothing to act on.

Four things can still break it.

• Using American-style options. In January 2019 a Robinhood user built a box on UVXY, an ETF with American-style options, declared it could not lose, and had short legs exercised early. Roughly $5,000 of capital became a deficit in the neighborhood of $57,000, and Robinhood subsequently stopped permitting the strategy.¹¹ Schwab lists the same failure mode: early assignment on a short leg “can effectively unravel the strategy,” exposing the trader to interest-rate, dividend and hard-to-borrow risk.³
• Closing early. You can buy the box back at any time, at the market. A short box behaves like a fixed-rate liability, so if rates have fallen since you sold it, buying it back costs more than you received, plus whatever the bid/ask takes. OCC states this from the seller’s side: “If interest rates fall, the value of the box increases, and closing the position early could result in a realized loss.”⁶
• Being liquidated. The box does not fail; your collateral does, and the broker unwinds positions at whatever the market offers.
• Timing at settlement. The amount is exact, the cash is not instantaneous. Settlement delivers cash the business day following expiration, so the obligation lands with a one-day lag you need to have funded.

One practical detail from OCC that appears almost nowhere else: box seller cash obligations aggregate at the clearing-firm level and concentrate on quarterly expirations, which can push a firm’s liquidity demand past OCC’s committed credit facilities. OCC runs a liquidity margin call policy starting 30 days before expiration, and advises that “to avoid the policy thresholds market participants may wish to look at using non-quarterly expiries as the basis for option boxes.”⁷

How the financing cost is taxed

SPX options are listed options that are not equity options, which makes them nonequity options under §1256(b)(1)(C) and therefore §1256 contracts. IRS Publication 550 defines nonequity options to include “broad-based stock index options” and names the Standard and Poor’s 500 index as an example.¹² SPY options are options on shares of an ETF, so they are equity options and fall outside this treatment entirely.

Three consequences follow. Open §1256 positions are treated as sold at fair market value on the last business day of the year. Gains and losses are 60% long-term and 40% short-term regardless of holding period. And §1256(c)(1) extends the same rules to the termination of an obligation, which matters for a short box whose legs are written rather than held.

So the financing cost arrives as a capital loss with 60/40 character. The timing is not smooth. A box open across December 31 is marked to market, so a single year’s recognized amount reflects that year’s move in the box’s value and can be larger, smaller or the opposite sign from the financing cost accrued in that year, with §1256(a)(2) truing it up when the position closes. Whether that is worth anything depends on the rest of your return. §1211(b) allows net capital losses against ordinary income only up to $3,000 a year, with the excess carried forward indefinitely under §1212(b). An investor generating $10,000 of box financing cost against $200,000 of realized gains has a genuinely useful deduction. An investor with no capital gains has a $3,000 deduction and a carryforward schedule.

There is a §1256-specific carryback at §1212(c), elected by checking box D on Form 6781, that carries a net §1256 loss back three years. It is allowed only to the extent of net §1256 contract gain in the carryback year.¹³ Somebody who borrows via boxes and trades nothing else in the §1256 world has no prior gains to carry back against, so the provision does nothing for them.

Compare that to margin interest. Investment interest is deductible under §163(d) only up to net investment income, with the excess carried forward, and Publication 550 is explicit that you must itemize on Schedule A and attach Form 4952 to claim it.¹² The shelter is narrower than it sounds. Investment income for this purpose covers interest and ordinary dividends, but Publication 550 excludes qualified dividends and net capital gain “unless you choose to include them,” and choosing costs you the preferential rate on the amount elected, dollar for dollar. For an investor whose dividends are mostly qualified, there is less to shelter than the headline suggests. The two regimes fail in opposite directions: margin interest needs itemized deductions and mostly offsets interest, while a box loss needs no itemizing but mostly offsets capital gains. Which is better is a fact about your return, not about the instrument.

Where the mortgage comparison breaks down

The most aggressive pitch for box spreads is home financing, and it is worth being precise about who is making it. Cboe’s Insights blog hosts a post titled “Long-Dated Box Spreads: A Better Way to Buy a Home,” bylined by Joseph Wang, cofounder of SyntheticFi, a firm that sells box-spread loans.⁹ A second Cboe-hosted post advocating boxes for borrowing and lending is bylined by the CEO of Alpha Architect, which sponsors the BOXX ETF. The post written by Cboe’s own staff reaches the opposite conclusion: short boxes “should be considered for short duration and clearly defined needs, not for long-term financing or lifestyle leverage.”⁸

Cboe’s market data supports its staff rather than its guest writers. Seventy-nine percent of SPX box contracts expire within six months, and the median traded loan is $100,000.² The long-dated segment that a mortgage substitute would need is a thin slice of the market.

Three numbers get conflated in this discussion and should be kept apart. Cboe’s listing rules permit LEAPS expiring 12 to 180 months out. What is actually listed runs to roughly five years, which is what the home-financing post uses.¹⁰ What is liquid is well under six months. Fifteen years of permitted listings is not fifteen years of financing.

Set the two instruments side by side at their real terms. A five-year box is a bullet obligation, collateralized by a portfolio marked to market daily, with the entire principal due at maturity and no contractual right to roll. A 30-year fixed mortgage amortizes, cannot be called while you pay it, is freely prepayable, and is indifferent to what the S&P 500 does. Matching maturities properly, the December 2031 box index sat at 4.64% on August 19, 2026 against a 6.67% thirty-year mortgage, so the box is cheaper by 203 basis points. It is also a different security with a different failure mode, and the rate gap is partly the price of that difference.

Then there is the refinancing question. If you cannot repay the box at year five from resources you already have, you are assuming you will sell another one. That assumes rates, your portfolio, your broker's margin rules, and box market liquidity all cooperate five years from now. The convenience-yield research says the spread you would pay quadruples in a crisis, which is the same environment in which your collateral would be down.

Where box spreads earn their complexity

Two cases hold up well.

Refinancing borrowing you already have. Somebody carrying a $250,000 margin balance at Schwab’s 10.075% has already accepted portfolio collateral, margin-call risk and the leverage itself. Replacing that balance with a box changes the financing terms without adding leverage: the rate falls by more than five points, the rate stops floating, and the cost becomes a capital loss that may be useful. The risk position does shift even so. Floating-rate exposure becomes a fixed rate with a bullet maturity to refinance, and the position picks up box mark and liquidity risk it did not have. The leverage decision was made previously, and this improves its terms.

A bridge with a repayment source that already exists. A home closing before another one settles, a contractually scheduled distribution, a vesting date on the calendar. Selling appreciated shares to cover a three-month gap triggers gains and exits the market; a short box costs a few hundred basis points annualized on a fraction of a year. The distinguishing feature is that repayment does not depend on the market rising.

Size matters more than the fee schedules suggest. Commissions are trivial at any realistic scale, a few dollars on a $100,000 box. What is not trivial is the bid/ask you cross on an off-the-run strike width, and the cost of a mistake. The market’s working unit is the 1,000-point box, which is 92% of volume, and Cboe’s median traded loan is $100,000.² Below roughly $25,000, the spread plus the risk of misconstructing four legs is likely to swamp what you save against an ordinary margin loan. That threshold is reasoning from the market-structure data rather than a published figure, so treat it as a rough floor.

When borrowing this way is a mistake

• Buying more equities with it. A 4.45% rate against a historical equity return looks like free money and is not. You have added a fixed obligation against a volatile asset, plus sequence risk and forced-liquidation risk. The box is a good way to arrange leverage and says nothing about whether to take it.
• Holding a concentrated position to avoid the tax. Borrowing to defer a capital gain leaves the concentration in place and adds a fixed obligation on top of the single-stock risk that prompted the question.
• Funding ordinary spending. A liability with a fixed maturity against an income stream that does not grow to meet it.
• Permanent retirement withdrawals. A retiree drawing on a box has to repay or roll it during exactly the drawdowns that make sequence risk dangerous.
• Replacing an emergency fund. Access depends on maintaining margin capacity, which is least available when markets are stressed and you most need the cash.
• Any plan whose repayment depends on the market rising. If the answer to “how do you repay this” is “the portfolio will be bigger,” the answer is missing.

An execution checklist

1. Use European, cash-settled index options. SPX is the liquid choice. American-style options on stocks or ETFs are the one error with a documented catastrophic outcome.
2. Verify all four legs share the same root, the same expiration and the same settlement style, and that the two strikes match in pairs. SPX and SPXW sit in one chain and can show the same calendar date while settling against different prints, so matching the date alone is not enough.
3. Submit the four legs as a single complex order with a limit price. Legging in creates unhedged directional exposure and a different margin requirement while you are half-filled.
4. Confirm the ticket shows a net credit smaller than strike width times $100 times contracts, and that the risk graph is flat.
5. Prefer 1,000-point width and strikes bracketing the index, where the volume is.
6. Compute the compounded ACT/360 rate before sending, then compare it against your own margin rate, your PAL, a HELOC and the cost of selling shares.
7. Consider a non-quarterly expiration, per OCC’s liquidity margin guidance.
8. Avoid an expiration in the last days of December, where the year-end mark and the settlement can land in different tax years.
9. Stress the collateral: a 50% equity decline, a house margin requirement raised without notice, no new income, and no ability to roll. If that forces liquidation, the borrowing is too large.
10. Name the repayment source before entering, and write down the deleveraging rule you will follow, since nothing on your statement will remind you the loan exists.

A note on who publishes what

Almost every well-ranking page on this topic is written by someone with a commercial interest in the answer. Cboe lists the contracts. Alpha Architect sponsors an ETF built on boxes. SyntheticFi, Vest and similar firms sell managed box borrowing. The market-structure data Cboe publishes is the best public source available on this instrument, and much of the rest of that material is accurate on the mechanics. The pattern to watch for is in the framing: the comparator chosen is usually the most expensive one available, and the $3,000 capital-loss limitation tends to go unmentioned.

The two unconflicted treatments worth reading are Kitces’ February 2026 piece, which places boxes in a tiered hierarchy alongside SBLOCs and HELOCs and states the capital-loss limitation correctly, and Harry Sit’s Finance Buff walkthrough, which is the clearest retail execution account in print. Cboe’s own staff piece is more cautious than either of the guest posts on its own site.

How Summitward helps

Borrowing against a portfolio changes the funded status of a plan, not just its cost line. A fixed obligation with a known maturity is a liability with a present value, and the assets backing it are the same assets a retirement projection assumes will be there.

Frequently asked questions

Is a box spread loan cheaper than a margin loan?

Almost always cheaper than a full-service broker’s standard margin schedule, which ran from 10.075% to 11.825% at Schwab in August 2026. Against a Pledged Asset Line at 6% to 8%, meaningfully cheaper. Against Interactive Brokers’ tiered margin at 4.38% to 5.13%, the difference is small enough that execution costs and a managed-service fee can erase it. Run the comparison against the rate you could obtain today.

Can I do this with SPY options instead?

No. SPY options are American-style options on ETF shares, so a short leg can be assigned early and unravel the structure, and they are equity options rather than §1256 contracts, so the tax treatment differs as well. The 2019 UVXY case is what this looks like when it goes wrong.

What happens if the market crashes while I have a box open?

The box is unaffected: it still settles to the strike width times $100. Your collateral is what falls, and the broker can issue a maintenance call and liquidate securities without notifying you first. The obligation stays constant while the assets backing it shrink, which is the mismatch Cboe identifies as the primary risk.

Do I need portfolio margin?

Not strictly, but Reg T treatment consumes buying power well beyond the economic risk of the position, which is why Cboe describes short boxes in Reg T accounts as capital intensive and points toward risk-based margin. Portfolio margin typically carries a six-figure account minimum and additional approval requirements.

How small can a box loan be?

Mechanically, very small: a 100-point width is $10,000 of obligation. Practically, off-the-run widths trade wider, and 92% of SPX box volume sits at 1,000 points. Below roughly $25,000 the spread you cross plus the risk of a construction error is likely to exceed what you save against an ordinary margin loan.

Can I pay it off early?

You can buy the box back at any time. There is no prepayment penalty, but you transact at the market: a short box behaves like a fixed-rate liability, so it costs more to repurchase if rates have fallen since you sold it, and you pay the bid/ask both ways.

Is the financing cost tax deductible?

It generally produces a §1256 capital loss, split 60/40 long and short term, marked to market at year end. That offsets capital gains without limit. Against ordinary income, §1211(b) caps net capital losses at $3,000 a year with the remainder carried forward. Vendor material describing the cost as fully deductible with no cap is describing only the first of those two situations.

Is BOXX the same thing?

BOXX holds the lending side, buying boxes to earn a T-bill-like return, which is the mirror of what a borrower does. Its tax treatment is the subject of the §1258 controversy discussed above. Holding it does not give you a box spread loan, and it gets its own treatment in the BOXX guide.

Key takeaways

• The savings figure is a comparison, so pick the right comparator. Against Schwab’s published margin schedule a box is worth several percentage points. Against Interactive Brokers’ own tiers it is worth a fraction of one, and a managed service fee can turn it negative.
• The rate depends on which convention you quote. Four conventions on the same trade can span 20 to 45 basis points, and OCC’s own primer uses two of them a few pages apart. The compounded ACT/360 figure is the one comparable to a margin loan.
• The instrument is exact; the collateral is not. Four European legs of the same product, settling against one SET print, produce a payoff known to the penny, while the portfolio securing it is marked to market every day and can be liquidated without notice.
• The tax benefit is conditional on having capital gains. §1256 gives 60/40 treatment and no wash-sale rules, but §1211(b) limits net capital losses against ordinary income to $3,000 a year, and the §1212(c) carryback reaches only prior §1256 gains.
• Cboe’s own data undercuts the mortgage pitch. Seventy-nine percent of SPX box contracts expire within six months, and Cboe’s staff advises against long-term financing use, while the long-dated home-financing case on its blog is made by a vendor selling the product.
• The strongest use is refinancing borrowing you already decided to do. Converting an expensive floating margin balance into a fixed-rate box improves the terms without adding leverage. Discovering box spreads and then finding a reason to borrow is a different transaction.

Related guides

• Personal Leverage: Margin, Leveraged ETFs, Lifecycle Theory, and Forced Selling covers the prior question: whether to borrow at all, and what a margin call does to a portfolio that has to sell into a decline.
• BOXX ETF Explained covers the lending side wrapped in an ETF, its deferral and capital-gain character, and the §1258 argument over whether that character holds.
• Where to Park Cash places SGOV, T-bills and money market funds into reserve layers.
• The Black-Scholes Equation develops the put-call parity relationship that makes a box a pure interest-rate instrument.
• Options Risk and Reward is the counterweight on retail options generally, and on §1256 treatment for directional trades.
• Debt Is a Time Machine frames borrowing as moving money across your own life, which is the question a cheap rate cannot answer for you.
• Portable Alpha and Return Stacking covers embedded leverage through funds, an alternative route to the same exposure without an options ticket.
• Four Things That Can Happen to a Capital Gain covers the gain-deferral logic that borrowing against appreciated stock is usually trying to serve.

Sources and method

1. Mandy Xu, Cboe, Week of 8/3/2026: The Fed Holds but Bond Volatility Breaks Higher, August 3, 2026. Reports that SPX box spread implied yield typically trades at an average premium of about 32bp over SOFR, that the premium has recently more than doubled to a multi-year high of 69bp on constrained dealer balance sheets, and that daily notional loan volume has more than tripled in three years to a record $1.7B. Cboe Insights
2. Cboe, SPX Box Spreads market statistics page, whose own note reads “All data are as of June 23, 2025, unless stated otherwise.” Box-specific figures: $1.2B average daily notional; $100k median traded loan notional; 92% of contracts at 1,000-point strike distance; 79% expiring within six months; 65% using 5000/6000 strikes; lending rates averaging about 26bp over 3-month Treasuries and ranging 6 to 44bp in the first half of 2025; a June 2025 snapshot of 4.19% for boxes against a 10.950% median broker margin rate, with broker rates for $100k loans ranging 5.83% to 11.075%. The page’s separate figure of 3.63 million contracts a day is total SPX options volume, not box volume, and is quoted here only to avoid the misattribution: $1.2B of box notional at $100k per 1,000-point box is roughly 12,000 boxes or 48,000 contracts a day. Cboe
3. Joe Mazzola, Charles Schwab, What Are Box Spreads?, November 20, 2025. Short box executed October 16, 2025 on SPX 6,600/6,700 expiring March 20, 2026: $9,830 credit, $10,000 settlement, $170 implied interest, about 4.15% annualized. Also the source for “this is a misconception” on riskless framing, the early-assignment warning for American-style options, and the net liquidation value warning. Schwab
4. Jules van Binsbergen, William Diamond and Marco Grotteria, Risk-Free Interest Rates, NBER Working Paper 26138, Journal of Financial Economics 2022. Rates inferred from put-call parity on European index options. Treasury convenience yield about 40 basis points on average, larger below three months maturity, quadrupling during the financial crisis. NBER
5. Federal Reserve Bank of New York, reference rates, August 17, 2026. SOFR 3.66%, effective federal funds rate 3.63%, target range 3.50% to 3.75%. Treasury par yields for the same date from the U.S. Treasury daily yield curve: 3-month 3.87%, 6-month 3.95%, 1-year 4.00%, 2-year 4.19%, 5-year 4.38%, 30-year 5.31%. NY Fed
6. The Options Clearing Corporation, Option Box Spreads for Investors, 2025. Worked example: $96,000 net debit against a $100,000 payoff one year out, rate computed as (1,000 − 960) / (960 × 365/365) = 4.16%. Also the source for the seller-side rate exposure (“If interest rates fall, the value of the box increases, and closing the position early could result in a realized loss”), the conclusion that box spreads are “best suited for institutional traders or experienced individuals,” and the disclaimer that commissions, fees, margin interest and taxes are excluded from its examples. OIC (PDF)
7. The Options Clearing Corporation, Box Spreads: Exchange-listed Options Strategies for Borrowing or Lending Cash, 2020. Source for the European-style rationale, the 46-day 999.35 bid / 999.40 offered example with rates of 0.48% and 0.52% (with the bid and offer labels transposed in the rate section), the “discount yield” characterization computed on a different basis than the box formula, the note that box spreads are “considered a strategy for professional traders and market-makers due to the transaction costs involved,” and the liquidity margin call policy footnote recommending non-quarterly expiries. OIC (PDF)
8. Rick Rosenthal, Director of Derivatives Sales, Cboe, SPX Box Spreads: What Every Advisor Should Know, March 12, 2026. Source for “Short box spreads should be considered for short duration and clearly defined needs, not for long-term financing or lifestyle leverage,” the asset-liability mismatch as the biggest risk, and the Reg T versus portfolio margin discussion. Cboe Insights
9. Joseph Wang, Cofounder, SyntheticFi, Long-Dated Box Spreads: A Better Way to Buy a Home, published on Cboe Insights, March 26, 2025. Source for the “fully tax-deductible as a capital loss” claim, the statement that Cboe offers SPX options with up to 5-year expiry, and the Reg T collateral example in which a $1.25MM box requires at least a $2.5MM portfolio. Cboe Insights
10. Cboe, S&P 500 Index Options Specifications. European exercise; cash settlement delivered the business day following expiration; $100 multiplier; up to twelve standard monthly expirations plus up to ten LEAPS expirations from 12 to 180 months; exercise-settlement value for standard SPX calculated from opening sales prices of the component securities on the expiration date, and from closing prices for PM-settled SPXW. Cboe
11. Box spread, Wikipedia, and contemporaneous coverage of the January 2019 incident in which a Robinhood user constructed a box on UVXY using American-style options and had short legs exercised early. Reported losses exceed $57,000 against roughly $5,000 of initial capital; figures differ across accounts, so the magnitude is given here as a range. Robinhood no longer lists box spreads among its permitted advanced strategies. Wikipedia
12. Internal Revenue Service, Publication 550, Investment Income and Expenses. Nonequity options include “broad-based stock index options,” where a broad-based index is “based on the value of a group of diversified stocks or securities (such as the Standard and Poor’s 500 index).” Also the source for the requirement to itemize on Schedule A and attach Form 4952 to deduct investment interest. Statutory provisions cited in the text: 26 U.S.C. §1256(a)(1), (a)(3), (a)(4), (b)(1)(C), (c)(1); §1211(b); §1212(b), (c); §163(d); §1092; §1258. IRS (PDF)
13. Internal Revenue Service, Form 6781, Gains and Losses From Section 1256 Contracts and Straddles. The net section 1256 loss carryback election is made by checking box D and entering the carryback amount on line 6, carried to the earliest of the three preceding years first, and limited to net section 1256 contract gain in each carryback year. The form’s instructions also state that the wash sale rules do not apply to section 1256 contracts. IRS (PDF)
14. Comparison rates, all as published in August 2026. Schwab margin: base rate 10.00% (last changed December 12, 2025), effective rates 11.825% under $25,000 down to 10.075% for $250,000–$500,000, with rates above $500,000 quoted on request. Schwab Pledged Asset Line, as of August 10, 2026: SOFR + 4.40% (8.03%) at $100,000–$250,000 down to SOFR + 2.40% (6.03%) above $2.5 million. Interactive Brokers Pro, USD benchmark 3.63%: 5.130% to $100,000, 4.630% to $1 million, 4.380% to $50 million. Bankrate national average HELOC 7.30% as of August 12, 2026. Freddie Mac Primary Mortgage Market Survey 30-year fixed 6.67% for the week ending August 13, 2026.
15. Cboe, Box Rate indices. Cboe publishes one index per listed SPX expiration, quoted in percent. Values used here are as of August 19, 2026: BOX20X26 (November 20, 2026) 4.16%, BOX18Z26 (December 18, 2026) 4.24%, BOX17U27 (September 17, 2027) 4.45%, and BOX19Z31 (December 19, 2031) 4.64%. Quotes are delayed and two nearer-dated series had stopped updating when this was checked, so confirm the series you intend to trade rather than assuming every listed index is live. These are indications of where the market has been pricing, not quotes you can lift. Cboe
16. Method. The calculator’s arithmetic lives in web/src/lib/box-spread-math.ts and is unit tested against OCC’s two published examples. The obligation is strike width × $100 × contracts, which is the settlement value of a European, cash-settled index box regardless of where the index prints. Four rate conventions are computed from the same inputs: simple ACT/365 over proceeds, bank discount ACT/360 over face, and internal rate of return on both ACT/360 and ACT/365 bases. Box rates are taken from Cboe’s per-expiration index matching the tenor under discussion, rather than by adding a spread to an overnight rate, because boxes have a term structure. Comparator rates that are charged in tiers are blended across those tiers for the amount borrowed. The collateral stress figure is 1 − (L/V)/(1 − m) for a loan L against collateral V with a required equity fraction m, which assumes the broker does not raise its house requirement and does not model how any particular broker margins a short box. Nothing here models commissions, exchange fees, bid/ask cost, or the possibility that a quoted midpoint will not fill. Comparison figures are first-year and undiscounted, and every alternative rate shown floats while a box’s cost is fixed to its maturity, so the comparison is a snapshot rather than a lifetime total. Nothing here is a forecast, and nothing here is tax, legal or investment advice.