Funding rate arbitrage: how the crypto basis trade works

Funding rate arbitrage pairs a spot purchase with an equal short in perpetual futures so that price moves cancel out and the funding payment becomes the return, and a year of exchange data shows how much that payment moves.

Funding rate arbitrage is a delta-neutral trade: buy a crypto asset in the spot market, short the same amount in perpetual futures, and collect the funding payments that longs pay shorts whenever the perp trades above spot. Price moves on the two legs offset each other, so the profit comes from funding, minus trading costs. On Binance's BTCUSDT perpetual, funding averaged 6.82% annualised over the 30 days to 24 September 2026, and 3.40% over the full year, with 23% of funding intervals negative.

The same trade in its dated-futures form is the crypto basis trade, also called cash and carry, and it's the engine behind synthetic dollars such as Ethena's USDe and the delta-neutral vaults that package it for passive holders. The mechanics are simple enough; the return depends on a rate that resets every few hours, and on how the book survives the days when it flips.

How does funding rate arbitrage work?

A perpetual future never expires, so exchanges keep its price close to spot with a periodic payment between the two sides. When the perp trades above the spot index, funding is positive and longs pay shorts; when it trades below, funding is negative and shorts pay longs. Binance settles BTCUSDT funding every 8 hours, while the onchain perps venue Hyperliquid settles every hour.

1. Find a positive rate. Look for funding that has stayed positive for several days, since a single high print often reverses at the next settlement.
2. Buy spot. Buy the asset for a set notional, for example $100,000 of BTC.
3. Short the perp. Open a short perpetual position for the same $100,000 notional. Net price exposure (delta) is now close to zero.
4. Collect funding. At each settlement the short receives funding on its notional while funding is positive, and pays it when funding turns negative.
5. Close both legs together. Sell the spot and buy back the short at the same time, so a price move between the two trades doesn't leave an open position.

Positive funding means the market is paying for borrowed long exposure, usually because traders are bullish. The short in this trade is the side getting paid, which is why funding income tends to be highest in rising markets and dries up, or turns into a cost, in falling ones.

A worked example with fees

The figures below use the actual 30-day average funding on Binance BTCUSDT to 24 September 2026, 0.00623% per 8-hour interval, and illustrative trading costs of 0.05% per trade. Fee schedules vary by venue and volume tier, so the fee line is an assumption for the example.

The capital involved is larger than the notional. Holding $100,000 of spot and margining a $100,000 short with one-third margin (3x) ties up about $133,333, so the 30-day average works out to roughly 5.1% a year on capital before costs. Thinner margin on the short leg raises the return on capital and brings the liquidation price closer, which is the trade-off every version of this strategy has to set.

What is bitcoin basis trading?

Basis trading is the dated-futures version of the same idea. A trader buys spot bitcoin and sells a future that expires on a set date at a higher price, locking in the gap, called the basis, which converges to zero at expiry. The return is known at entry, which is the main difference from funding, where the payment resets every settlement.

On 24 September 2026 Deribit's December 2026 bitcoin future traded 1.26% above the BTC index, about 5.0% annualised over 92 days, and the September 2027 contract traded 5.27% above, about 5.3% annualised. The same structure exists in traditional markets: the "Treasury basis trade" that dominates search results for the term pairs cash US Treasuries with Treasury futures, run by hedge funds with repo financing. The crypto version is smaller and carries crypto venue risk, though the arithmetic of buying the cheaper leg and selling the richer one is identical.

How Ethena turned the trade into a synthetic dollar

The idea of packaging funding rate arbitrage as a dollar token traces back to a March 2023 essay by BitMEX co-founder Arthur Hayes. In Dust on Crust, published on 8 March 2023, he proposed the "Satoshi Nakamoto Dollar", defined as "$1 of Bitcoin + Short 1 Bitcoin / USD Inverse Perpetual Swap." The long and the short cancel out, so the position holds a steady dollar value, and Hayes noted that such swaps "historically have net paid interest to shorts," which meant the structure would earn funding on top of holding its value.

Ethena built that design into USDe, and its own genesis story credits the essay directly. USDe mostly uses ETH and staked ETH rather than BTC, pairing spot collateral with short perpetual positions on centralised exchanges, and it passes the funding and staking income to holders who stake into sUSDe. On 24 September 2026, sUSDe paid 4.67% with a 30-day average of 4.73%, and USDe supply stood at $4.9 billion, down from a peak of $14.8 billion on 4 October 2025, according to DefiLlama.

Supply has moved with the trade's fortunes. It nearly tripled from $5.3 billion on 1 July 2025 to $14.7 billion on 1 October 2025, fell by half within two months of the 10 October liquidations, and dropped from $5.9 billion to $3.9 billion over April 2026, after funding had run negative since February. USDe also briefly traded at about $0.65 on Binance's own order book during the 10 October 2025 liquidations while holding near $1 elsewhere, an episode covered in what happens when a stablecoin depegs.

What funding has paid over the last year

Funding is the whole return, so its history is the most useful number for anyone sizing this trade. The table shows monthly averages on Binance's BTCUSDT and ETHUSDT perpetuals, annualised from 8-hour settlements, pulled from Binance's public funding-rate API on 24 September 2026.

Two patterns stand out. From February to April 2026 funding ran negative on both assets for three straight months, so a short perp paid out; a $100,000 book lost about $333 in funding over that stretch before any fees. The ceiling was also low: neither contract printed above 0.01% per 8 hours at any point in the 12 months, a fifth of the +0.05% rate often used in textbook examples, so the double-digit annualised figures quoted in older guides didn't appear on Binance in this period.

Does funding rate arbitrage still work in 2026?

Yes, at a modest rate that depends on market direction. Over the 12 months to 24 September 2026, BTC funding on Binance averaged 3.40% annualised and ETH 2.47%, against 3.62% paid to USDC suppliers on Aave's Ethereum market on 24 September 2026. The more recent window is stronger: 6.82% for BTC over the last 30 days, with a single negative interval.

The trade works best in sustained rallies with heavy long demand, and it can run at a loss for months in a falling market, as February to April 2026 showed. Fees and the cost of idle margin decide whether a positive rate turns into a net return, which is why this is mostly a business for desks and vault curators with low trading costs and the ability to rotate capital elsewhere when funding flips.

Running the trade across two venues

Funding differs by venue, which opens a second version of the trade: short the perp where funding is highest and hedge where it's lowest, or against spot held elsewhere. Over the same 30 days to 24 September 2026, BTC funding on Hyperliquid averaged 9.32% annualised against 6.82% on Binance, and ETH averaged 9.93% against 4.72%. Most of that gap comes from Hyperliquid's default rate of 0.00125% per hour (about 11% annualised), which BTC funding sat at in 519 of the last 720 hourly settlements.

A cross-venue spread adds its own risks. The two legs sit in separate margin accounts, so a sharp move can push one short toward liquidation while the offsetting gain is stuck on another venue, and moving collateral between them takes time. Rates also converge: on the afternoon of 24 September, Binance BTC funding had dropped to about 1.6% annualised and OKX and Bybit were printing slightly negative, while Hyperliquid stayed at its default.

The risks in funding rate arbitrage

• Rate flip: funding can turn negative at the next settlement, and did so for most of February to April 2026. The short then pays instead of collecting.
• Basis risk: spot and perp prices can drift apart. If the perp trades at a discount when the position is closed, the exit costs more than the funding collected.
• Liquidation of the short leg: the short is a margined position. At 3x, a 33% rally wipes out the margin posted against it, and exchanges liquidate earlier, at their maintenance margin. The spot leg gains the same amount, but if the gain can't reach the short's margin account in time, the short is closed and the book is left long and unhedged.
• Auto-deleveraging: when insurance funds run dry, exchanges force-close profitable positions to cover losing ones. On 10 October 2025, more than $19 billion of margined futures positions were liquidated in about a day and several venues, Hyperliquid included, force-closed winning shorts through auto-deleveraging, which broke delta-neutral books by removing the hedge mid-crash (CoinGecko).
• Venue and custody risk: collateral held on an exchange is exposed to that exchange's solvency. Collateral held onchain is exposed to the venue's contracts and bridge.
• Costs: trading fees on four trades, spreads, and the opportunity cost of margin can exceed funding when rates are low, which they were for much of the last year.

Always make sure to do your own research and be aware of the above and any other risks before depositing.

How a delta-neutral vault packages the trade

Most holders don't want to run two margin accounts and watch funding every hour, so the trade is increasingly sold in packaged form. A synthetic dollar such as USDe is one packaged form, with the issuer running the hedge for every holder at once. A delta-neutral vault does the same job with a named curator running the book: depositors put in stablecoins, receive vault shares, and the curator runs the spot and short legs within the vault's mandate.

On Upshift, the Axis Origin vault gives depositors access to Axis's market-neutral arbitrage book and showed a trailing 30-day APY of 8.7% on 24 September 2026 (Upshift data). Upshift vaults are non-custodial: the curator trades through a subaccount restricted by a policy engine to approved chains, protocols, tokens and functions, and can't send funds anywhere outside that perimeter.

Where a vault's return comes from

• Funding on the short leg: the core of the return, and the part that can turn negative.
• Basis: when the book uses dated futures, the gap between futures and spot, locked at entry.
• Yield on the collateral: spot ETH can be held as staked ETH, and stablecoin margin can sit in a lending market, adding a second return under the same hedge.
• Rotation: when funding flips negative, a curator can shrink the book and move capital into lending until rates recover, which a single-strategy product can't.

A worked vault example

The table uses an illustrative $10 million vault. The funding rates are real, the allocation, costs and fee are assumptions for the example, and none of it describes a specific Upshift vault.

Swapping in other funding windows shows how much the rate drives the outcome. Gross funding on the same vault would have been 2.55% of the vault at the 12-month Binance average of 3.40%, 6.99% at Hyperliquid's 30-day average of 9.32%, and -1.02% a year at the February to April 2026 average of -1.37%. A curator that cut the book during those three months and lent the stablecoins instead would have avoided most of that drag, which is the main case for a managed vault over a static position.

A vault adds its own layer to the risks above: smart contract risk in the vault, the curator's execution, and redemption timing, since both legs have to be unwound before capital comes back. Upshift's contracts have been through 10 smart contract audits by 6 independent firms, and they don't carry smart contract insurance. Each vault shows its withdrawal terms before deposit.

Frequently asked questions

What is funding rate arbitrage?

It's a trade that buys an asset in the spot market and shorts the same amount in perpetual futures, so price moves cancel and the position earns the funding payments longs make to shorts when funding is positive.

Is funding rate arbitrage profitable?

It has been, at modest rates. Binance BTC funding averaged 3.40% annualised over the year to 24 September 2026 and 6.82% over the last 30 days, before trading costs. It lost money from February to April 2026, when funding stayed negative.

What happens if funding is positive?

Longs pay shorts at each settlement, in proportion to position size. A delta-neutral trader holding the short collects that payment.

What happens if funding turns negative?

Shorts pay longs, so the trade starts costing money. Traders either hold through it, reverse the legs, or close the position and wait.

Is basis trading the same as funding rate arbitrage?

They're close relatives. Basis trading uses dated futures and locks in the premium at entry; funding rate arbitrage uses perpetual futures, where the payment resets every settlement.

Can funding rate arbitrage be done onchain?

Yes. Onchain perps venues such as Hyperliquid pay funding hourly, and vaults and synthetic dollars run the trade for depositors who hold only a stablecoin.

Keep reading

• The DeFi yield strategies onchain vaults run. Basis trades alongside lending, looping and RWA credit.
• How to earn yield on bitcoin. Where the BTC basis trade fits among lending and BTC vaults.
• What are DeFi yield vaults. How to evaluate a vault and what its share token represents.