Crypto Cash-and-Carry Arbitrage: Spot–Futures Basis, Net Carry and Risks

Editorial responsibility: Oliver Sinz · Last verified: October 14, 2026

Crypto cash-and-carry arbitrage means buying bitcoin in the spot market and selling a dated futures contract that trades above spot. If both positions survive to expiration and the futures contract settles against a price at which the spot position can actually be sold, the price moves on the two legs cancel. What remains is roughly the premium that existed when the trade was opened.

"Roughly" does a lot of work in that sentence. The premium is a gross number. Between it and the money that ends up in an account sit trading fees, slippage, the cost of financing the bitcoin, idle collateral, the settlement rules of one specific contract, and the requirement that the short futures position pass every margin check until expiry. This guide works through that accounting with three fully calculated examples. Every price, rate and premium in them is hypothetical, not current market data.

The two legs and the premium

This guide defines the basis as futures minus spot: B = F − S. A positive basis means the futures contract trades at a premium. Some sources, including Binance Academy, use spot minus futures instead, so always check the sign before comparing figures.

The trade buys Q bitcoin at spot price S0 and sells futures on the same Q bitcoin at F0. For a linear, USD-settled contract, the profit and loss at any later time t is:

• Spot leg: Q × (S_t − S0)
• Short futures leg: Q × (F0 − F_t)
• Combined: Q × [(F0 − S0) − (F_t − S_t)] = Q × (B0 − B_t)

The combined result depends only on how the basis changed, not on where bitcoin went. At expiration, the futures contract settles at a final settlement price R. If the spot bitcoin is sold at exactly R, then F_T = S_T = R, the terminal basis is zero, and the combined result is Q × B0.

Two details matter. First, the premium is not paid to you when you open the short. Selling a future at a high price brings in no cash; the premium is earned gradually through mark-to-market gains and losses as the futures price converges. Second, the identity only holds if both legs exist in matched size until expiry and are valued at the same reference price. Examples 2 and 3 below show what happens when those conditions fail.

Sizing the hedge: match bitcoin, not contracts

The futures quantity is the bitcoin quantity divided by the contract unit. CME's standard Bitcoin futures represent 5 BTC per contract, and its Micro Bitcoin futures represent 0.1 BTC; both are linear and cash-settled in U.S. dollars. A 10 BTC spot position therefore matches two standard contracts. A 12 BTC position would need two standard contracts plus 20 micros, because whole-contract sizes rarely divide evenly into an inventory.

Contract counts mean nothing across venues. "Two contracts" may be 10 BTC at one exchange and $20 of face value at another. Inverse contracts add a further wrinkle: Deribit's BTC futures have a face value of $10 per contract and settle in bitcoin, so a fixed number of contracts covers a changing amount of BTC as the price moves, and a hedge sized once at entry drifts. The examples here use linear contracts to keep the arithmetic clean.

Settlement decides what "convergence" means

Convergence is a contract term, not a law of nature. Each futures contract defines its own final settlement price, and venues define it differently:

• CME Bitcoin futures are cash-settled to the CME CF Bitcoin Reference Rate (BRR) on the last trading day, the last Friday of the contract month; trading ends at 4:00 p.m. London time (CME Rulebook, Chapter 350). The BRR is built from trades on constituent spot exchanges between 3:00 and 4:00 p.m. London time, split into twelve five-minute partitions and computed as the equally weighted average of each partition's volume-weighted median (CME).
• Deribit futures are cash-settled at a time-weighted average of the Deribit index between 07:30 and 08:00 UTC; inverse contracts settle in BTC and linear contracts in USDC.
• KuCoin's coin-margined BTC quarterlies settle in cash against a 30-minute TWAP of its spot index at 08:00 UTC and carry a 0.025% delivery fee, according to KuCoin's delivery rules (last updated July 2025).
• Physically delivered bitcoin futures also exist, for example at Bitnomial.

With cash settlement, the futures leg ends as a cash adjustment and the spot bitcoin must be sold separately. Any gap between the spot sale price and the reference rate stays in the result: Residual = Q × (S_exit − R). Selling 10 BTC at $107,900 when the reference settles at $108,000 costs $1,000. Traders can work the spot sale through the reference window or deal with OTC desks at the reference price, and CME's Basis Trade at Index Close (BTIC) lets participants trade futures at an agreed spread to a reference rate. Even "aligned" setups can differ: CF Benchmarks notes that most U.S. spot bitcoin ETFs strike their NAV off the BRR's New York variant, while CME futures settle to the London BRR — a different hour.

With physical delivery, the short delivers the bitcoin it holds and is paid the final settlement price, so there is no separate spot sale and no reference mismatch. In exchange, the bitcoin must be in the right account by the delivery deadline under that clearinghouse's procedures, and any delivery fees apply.

Example 1: a complete trade held to expiration

Basis. B0 = $102,000 − $100,000 = $2,000 per BTC, or 2.00% of spot. Simple annualized basis: 2.00% × 365 / 91 = 8.02%.

Gross P&L. Spot: 10 × ($108,000 − $100,000) = +$80,000. Short futures: 10 × ($102,000 − $108,000) = −$60,000. Combined: +$20,000, exactly 10 × $2,000. Bitcoin rose 8%, but the gain on one leg and the loss on the other cancel because the futures settled at the same price the spot was sold at.

Costs. Entry: spot fee $1,000, spot slippage $500, futures fee $204, futures slippage $510 = $2,214. Exit: spot fee $1,080, spot slippage $540, settlement fee $216 = $1,836. Total: $4,050.

Financing. $500,000 × 7.00% × 91 / 365 = $8,726.03.

Net profit. $20,000 − $4,050 − $8,726.03 = $7,223.97.

Capital committed. The trader's own money at inception is the unfinanced spot ($500,000) plus futures collateral ($300,000): $800,000. Entry costs are treated as expenses, not capital. Holding-period return: $7,223.97 / $800,000 = 0.903%. Simple annualized (ACT/365): 0.903% × 365 / 91 = 3.62%.

The cash-flow ledger separates what is an expense from what is merely money moving between accounts:

During the trade, futures account equity falls from $300,000 to $240,000, still above the $162,000 maintenance requirement at expiry. Under these margin assumptions, maintenance is breached if the futures price ever trades above about $114,783, a 12.5% rise from the $102,000 entry price. Example 1 assumes it never does. Example 3 drops that assumption.

Return measures that are easy to confuse

Borrowing lowered the return here. Leverage only raises return on equity when the trade's unlevered return exceeds the borrowing rate, and 4.92% is below 7.00%. Borrowing always does one thing reliably: it increases exposure to margin and funding stress. The right benchmark for either figure is what the same capital would earn in a short-term cash instrument, which depends on the account arrangements.

Annualized basis is not an annualized return

Quoting the premium as 8.02% a year is convenient for comparing contracts with different maturities. It is not a return forecast, for four reasons:

1. Different denominator. The basis is measured against spot notional. A trader's return is measured against their own capital, which includes futures collateral and excludes borrowed money.
2. Gross, not net. Fees, slippage, settlement charges and interest are not in it.
3. It assumes repetition. Annualizing a 91-day result implies the trader can roll into a similar premium every quarter. Premiums are not stable: CF Benchmarks reports the CME front-month annualized basis approached 25% in February 2024 and briefly dipped below zero in early 2025.
4. Conventions vary. Simple and compounded annualization give different numbers, as do 360- and 365-day years. CF Benchmarks' own worked example annualizes the basis simply but a monthly return with compounding. Always state which method is used.

Example 2: exiting before expiration

Same entry as Example 1, but both legs are closed on day 45. The relationship from the first section, applied per bitcoin, is Combined gross P&L = B0 − B1. It holds exactly for matched linear exposure when both legs are valued at their actual exit prices.

In both scenarios bitcoin rose, and in both the directional exposure was neutralized. What was left was basis exposure. In the widening case the futures premium grew from 2.0% to about 3.2% of spot during a rally, and the trade lost money even though nothing went wrong with the hedge ratio. Convergence at expiry would have reversed that $15,000 mark-to-market loss, but only for a trader who could keep both legs open. An exit forced by margin, financing, risk limits or a better use of capital turns a temporary basis move into a realized one.

Example 3: the hedge is fine, the futures account is not

Now suppose bitcoin rallies hard around day 30 over a few sessions. The futures account's equity is $300,000 − 10 × (F − $102,000), and under the illustrative 15% maintenance rule it first falls short when the futures price passes about $114,783. Assume the trader cannot meet that call before prices reach the following levels:

Economically, the position has lost $8,000. The futures account, however, needs $213,600 of new collateral, 26.7% of the original $800,000. The $250,000 spot gain sits in a different account, possibly at a different firm, and is partly financed by a loan. It does not flow to the futures account on its own.

That is collateral fragmentation. CME's acceptable-collateral schedule covers cash, government and agency securities, gold, select stocks and ETFs and other specified assets, mostly with haircuts; bitcoin itself is not on it. CME also states that a margin call restores the account to initial margin, and that margins tend to rise when volatility rises, which can enlarge a call exactly when prices are moving fastest. Some crypto venues let spot holdings count as collateral in a unified account after a haircut — Bybit applies a "collateral value ratio," with 0.98 for BTC in its own example — but that only helps if both legs sit on the same venue in the right account mode, and it concentrates exchange risk. The remaining options take time: wire new cash, borrow more, or sell some bitcoin, move the proceeds and cut the futures in proportion.

Assumed close-out rule (an illustration, not any venue's actual procedure): the unmet call results in the entire short being closed at $128,100, $300 above the prevailing futures price, plus a 0.02% fee. Realized futures loss: 10 × ($102,000 − $128,100) = −$261,000. Fee: $256.20. Collateral returned: $38,743.80 (the $300,000 posted, less the loss and the fee). Real procedures differ widely, from broker-managed margin calls at regulated futures commission merchants to automated, sometimes partial, liquidation engines at crypto exchanges.

• Outcome A — spot sold the same day at $125,000: +$250,000 spot gain, −$261,000 futures loss, −$256.20 close-out fee, −$1,875 spot exit costs, −$2,214 entry costs, −$2,876.71 in 30 days' interest. Net: −$18,221.91, against +$7,223.97 had the trade survived to expiry as in Example 1.
• Outcome B — spot kept unhedged, and bitcoin falls back to $100,000 by day 45 before it is sold: net −$269,285.27. With the short gone, every $1,000 move in bitcoin changes the result by $10,000. The "market-neutral" trade has become a leveraged long.

Reverse cash-and-carry

When futures trade below spot, the mirror trade borrows bitcoin, sells it spot, and buys the dated future. Held to expiry under the same idealized conditions, it earns Q × (S0 − F0) before borrow fees and trading costs. In practice it is harder to run. The bitcoin has to be borrowable in size for the entire term. Borrow rates float — Binance Margin, for example, updates rates and charges interest hourly — and lenders can restrict borrowing. The short spot position faces its own margin call if bitcoin rises, the fragmentation problem in reverse. Short-sale access is also limited for many retail accounts, depending on jurisdiction and venue. CF Benchmarks describes backwardation in the CME bitcoin market as relatively rare and short-lived, and when discounts do appear, few participants are positioned to arbitrage them.

Exchange, counterparty and operational risk

A hedge held at a failed venue is not a hedge. The collapse of FTX, where the SEC alleged that customer funds were diverted, showed that collateral and spot inventory can become inaccessible at once. Splitting legs across two venues doubles the points of failure and adds transfer risk: deposits and withdrawals can be suspended at the moment collateral needs to move. Other risks include outages near a settlement window, changes to index constituents or methodology, a missed roll, and simple execution errors when legging into or out of the trade.

Why the premium persists: limits of arbitrage

If the trade were truly riskless and costless, capital would compete the premium down to the cost of financing. It does not, and the literature explains why. Shleifer and Vishny's "The Limits of Arbitrage" (Journal of Finance, 1997) argues that real-world arbitrage requires capital and carries risk, and that it can become ineffective in extreme conditions, when prices stray furthest from fundamental value. For crypto specifically, a BIS working paper finds that crypto carry has averaged above 10% a year and at times exceeded 40%, that interest-rate differences explain little of it, and that it reflects leveraged demand from smaller, trend-chasing investors combined with arbitrage capital held back by regulatory and margin frictions. The authors also find that high carry tends to precede price crashes.

Example 3 shows the mechanism at the scale of a single trade: the rallies that widen the premium are the same rallies that drain short-futures accounts.

How this differs from spot–perpetual funding trades

Many crypto guides label long-spot, short-perpetual positions "cash and carry," but the economics differ. Perpetual futures have no expiry, so there is no contractual settlement that pulls the price to spot on a known date. The hedger's return comes from funding payments that change every interval and can flip sign, rather than from a premium fixed at entry and released through convergence. Dated-futures cash-and-carry carries basis risk until expiry; a perpetual hedge carries funding risk for as long as it stays open.

What to check before treating a premium as an opportunity

• Which basis sign convention the quote uses, and whether it is annualized simply or compounded.
• The contract unit, linear or inverse structure, settlement currency, and the exact final settlement reference and window.
• Whether you can sell spot at, or deliver against, that reference.
• All-in costs to open and close both legs, including any settlement or delivery fee.
• Financing cost compared with the trade's unlevered return.
• How large a rally the futures collateral can absorb, and how fast spot value can be turned into eligible collateral.
• Where each leg is held, and what happens if that venue stops processing withdrawals.

A positive futures premium is a starting point for analysis, not a yield. Whether it becomes profit depends on settlement terms, costs, funding, and the ability of both positions to survive the path to expiration.