8 Crypto Arbitrage Strategies in 2026, Ranked by How Long the Spread Lasts.

Crypto arbitrage spreads in 2026 range from a 0.035% gap between six major bitcoin exchanges, smaller than the fee to trade on any of them, to perpetual futures funding rates that BitMEX found were positive for more than 92% of the third quarter of 2025. The first kind closes before most traders see it. The second can be collected for months.

That difference, how long a spread survives once someone starts trading against it, decides which crypto arbitrage strategies an individual can realistically run and which belong to firms with servers next to the exchange. Most guides list arbitrage types without it. This one ranks eight strategies by it, with a worked example for each.

How these crypto arbitrage strategies are ranked

Each strategy is ranked by how persistent its spread has been in the most recent data available, from spreads that held for most of a quarter to spreads that last seconds. Every entry gives the spread, a worked example on a stated amount, what closes the spread, and who can run the trade. Figures are dated and attributed by source in the text. Spreads in crypto move daily, so treat every figure as a reading at a point in time, not a forecast.

The 8 crypto arbitrage strategies, ranked

1. Funding-rate arbitrage

The spread. Perpetual futures have no expiry, so exchanges keep their price close to spot with a funding payment that one side pays the other at fixed intervals, every eight hours on many venues. When the perpetual trades above spot, longs pay shorts. BitMEX's Q3 2025 derivatives report found funding was positive for more than 92% of that quarter, and that bitcoin funding sat at exactly 0.01% per interval for 78.19% of it, because the funding formula has an interest component that anchors the rate there.

Working example. Buy $10,000 of bitcoin spot and short $10,000 of the bitcoin perpetual. The two legs cancel each other's price exposure, which makes the position delta-neutral. At 0.01% per eight-hour interval, the short collects $1.00 per interval, $3.00 per day and $1,095 per year: 10.95% of the $10,000 hedged, before fees. A 0.10% taker fee on each side of the spot leg costs $20 of that.

What closes it. Crowding and sentiment. BitMEX's State of Crypto Perpetual Swaps 2025 report described funding-rate arbitrage as increasingly crowded, and on 16 April 2026 bitcoin's funding rate reached its most negative level since 2023, which means shorts were paying longs. A position that stays open through a negative stretch pays funding instead of collecting it.

Who can run it? Anyone with accounts on both a spot venue and a perpetuals venue, enough margin on the short to survive a sharp rally without liquidation, and time to rebalance the two legs as prices move. The alternative is a managed product. VectorTraders runs systematic, multi-strategy trading on crypto perpetual contracts, and its entry-level products are driven by funding-rate and basis arbitrage: clients deposit, the firm runs the hedged positions and rebalancing, and leverage is moderate and set by the firm. Its higher-yield products run quantitative contract trading, which has a different risk profile from the entry tier. Target returns are quoted net of management and performance fees, trading commissions and slippage. VectorTraders accepts US clients and serves clients in 90+ countries.

2. Cross-venue funding arbitrage

The spread. The same contract can carry different funding rates on different exchanges, because exchanges differ in margin design and in who trades on them. BitMEX's Q2 2026 derivatives report measured an average annualised funding premium of 7.17% for bitcoin perpetuals on Hyperliquid over Binance between 2023 and 2026, and 5.31% for ether perpetuals.

Working example. Short $50,000 of the bitcoin perpetual on the venue paying the higher rate and go long $50,000 on the venue paying the lower rate. At the 7.17% annualised average gap, the pair earns $3,585 a year on $50,000 of notional per leg, before fees, with no spot purchase needed. Both legs need their own margin, so the capital tied up is larger than either leg's margin alone.

What closes it. The gap is not constant. BitMEX advised traders to establish whether a gap is structural or event-driven before committing capital, because event-driven gaps reverse. Running two exchanges also means two counterparties, and a sharp move can liquidate one leg before the other is adjusted.

Who can run it? Traders are already active on both venues, including at least one decentralised exchange, with systems to monitor margin on both legs.

3. Cash-and-carry arbitrage (futures basis)

The spread. Dated futures usually trade above spot, and the gap, called the basis, shrinks to zero at expiry. On 7 August 2026, CME's August bitcoin contract settled at $65,175 against a New York bitcoin benchmark of $64,880, a gross annualised basis of 7.89%, according to CryptoSlate. The September contract showed 6.25% and the December contract 5.69%, against a two-year Treasury yield of 4.19% the same day.

Working example. Buy one bitcoin's worth of spot exposure at $64,880 and sell the August future at $65,175. The $295 difference, 0.4547% of the $64,880 spot price, is locked in at entry and collected at expiry whichever way bitcoin moves, which is how a short contract period annualises to 7.89% gross.

What closes it. Competition from institutions running the same trade through spot ETFs, and the alternative of holding Treasuries. Crypto.news reported that by mid-2026 the three-month CME basis had fallen to 3%, below the 3.8% two-year Treasury yield, at which point the trade earned less than a government bond.

Who can run it. Accounts with access to regulated futures and a spot ETF or spot holdings, enough capital for the contract size, and a funding cost below the basis.

4. Regional premium arbitrage (the kimchi premium)

The spread. Bitcoin sometimes trades at a different price on South Korean exchanges than elsewhere, a gap known as the kimchi premium. CryptoQuant's index showed a premium as high as 8.27% in October 2025, and Upbit traded 1% above Binance on 1 September 2026, according to Bloomberg data reported by ChainCatcher.

Working example. At a 1.00% premium, buying $100,000 of bitcoin abroad and selling it in Korea yields $1,000 gross, before exchange fees, transfer fees and the cost of converting Korean won back to dollars.

What closes it. Capital controls keep it open for residents and closed for everyone else: Korean exchanges require residency-based identity checks, and moving won out of the country is restricted. The premium also reverses. Four Pillars counted 134 days of discount out of 236 in 2026, and the Korea Herald reported that 2026 was the first year since CryptoQuant began publishing the data in July 2020 that discount days outnumbered premium days.

Who can run it? In practice, Korean residents with accounts both in Korea and abroad. For everyone else the kimchi premium is a market indicator, not a trade.

5. Cross-exchange spot arbitrage

The spread. Buying on the cheaper exchange and selling on the dearer one. At 13:18:06 UTC on 10 September 2026, CoinPaprika recorded Binance at $76,842.00 and Coinbase at $76,814.82, with six major venues quoting bitcoin within $27.18 of each other, a spread of 0.035%.

Working example. On $10,000, a 0.035% gap is $3.50. Buying on one exchange and selling on another at a 0.10% taker fee per trade costs $20 before any withdrawal fee, so the trade loses $16.50. Coinbase's base-tier taker fee of 0.60% widens that loss further.

What closes it. Market makers who hold balances on every major exchange and close gaps without moving coins between them. Gaps wider than fees appear mainly in stress; CryptoSlate reported a sustained gap between Coinbase and Binance during selling in January 2026.

Who can run it? Firms with pre-funded balances on several exchanges and low fee tiers. Software sold as a crypto arbitrage bot for retail traders usually competes for the same 0.035% against those firms.

6. DEX–CEX arbitrage

The spread. Decentralised exchange prices update only when someone trades on-chain, so they lag centralised exchange prices. A study by Wu, Sui, Thiery and Pai covering August 2023 to March 2025 estimated that 19 major searchers extracted $233.8 million from 7,203,560 CEX–DEX arbitrages on Ethereum, on $241.7 billion of volume.

Working example. Those figures put the average extracted value at $32.46 per arbitrage, or 0.0967% of the volume traded. A searcher pays part of that to block builders for priority inclusion in the block, so the amount kept is lower.

What closes it. Concentration. The same study found that by the first quarter of 2025, three searchers captured 73% of extracted value and 90% of arbitrage volume, and that profitability depended on close relationships with block builders.

Who can run it? Specialist trading firms with exchange balances, on-chain infrastructure and builder relationships. It is not open to individual traders in practice.

7. Stablecoin depeg arbitrage

The spread. A stablecoin trading below $1 on one venue while holding near $1 elsewhere. During the liquidation cascade of 10 October 2025, Ethena's USDe fell to $0.65 on Binance's order book between 21:42 and 21:51 UTC, according to Galaxy Research, while it barely moved on Curve.

Working example. Buying 10,000 USDe at $0.65 costs $6,500. Selling the same 10,000 USDe at $0.99 elsewhere returns $9,900, a gross gain of $3,400, or 52.3% of the $6,500 spent, if the tokens can be moved off the exchange before the gap closes.

What closes it. Time and access. The trough lasted nine minutes, withdrawals during a market-wide cascade are slow, and the same event liquidated traders who held USDe as collateral. A gap this size depends on a market failure, and market failures cannot be scheduled.

Who can run it? Traders with idle dollars already on the affected exchange and the ability to act within minutes. It cannot be planned as a source of regular income.

8. Triangular arbitrage

The spread. A mispricing across three trading pairs on one exchange, for example bitcoin to ether, ether to USDT and USDT back to bitcoin, where the round trip ends with more bitcoin than it started with.

Working example. Three trades at a 0.10% taker fee cost 0.30% of the amount traded, $30 on $10,000. The mispricing across the three pairs has to exceed 0.30% before the round trip earns anything.

What closes it. The exchange's own market makers and automated traders, who see the same order books with lower fees and faster connections. Because all three legs sit on one venue, the mispricing is visible to everyone at the same moment.

Who can run it? Automated traders on the lowest fee tiers with direct market access. It is the most common subject of crypto arbitrage bot tutorials and the least available to anyone running one from home.

What the ranking shows about crypto arbitrage

The crypto arbitrage spreads that last longest, funding and basis, are the ones that pay a counterparty for holding a position over time rather than for being faster. The spreads that close in seconds pay for speed and infrastructure, and most of that value goes to a small number of firms. For an individual, the practical question is not which spread is widest today, but which one lasts long enough to be collected after costs.

Crypto arbitrage FAQ

What is crypto arbitrage?

Crypto arbitrage is buying and selling the same asset, or two closely linked assets, in different places or forms to profit from a price difference between them. The two positions offset each other, so the profit comes from the gap closing rather than from the asset's price moving.

Is crypto arbitrage still profitable in 2026?

Some forms are and some are not. Cross-exchange spot gaps of 0.035% were smaller than a single 0.10% trading fee in September 2026, while funding and basis spreads have stayed positive for long periods. The CME basis was 7.89% gross annualised on 7 August 2026 but had fallen to 3% in mid-2026, below the two-year Treasury yield.

Is crypto arbitrage legal?

Arbitrage itself is a normal market activity, but each exchange's terms and each country's rules still apply. South Korea's capital controls and residency-based identity checks, for example, are why foreigners cannot trade the kimchi premium. This is not legal advice; check the rules where you live.

Do crypto arbitrage bots work?

A bot can execute a strategy, but it does not create a spread. For the fast strategies, cross-exchange, triangular and DEX–CEX, retail bots compete with firms that have lower fees, pre-funded balances and better connections. In the DEX–CEX study cited above, three searchers took 73% of the value.

How can I do crypto arbitrage without trading myself?

Through a managed product that runs the strategy for you. At VectorTraders, entry-level products are driven by funding-rate and basis arbitrage, and each plan's target net returns, minimum deposit and lock period are listed on the firm's fund plans page. Targets are estimates, not guarantees, and past performance does not guarantee future results.