Practical Guide to Crypto Funding Arbitrage

1. Introduction

Funding rate arbitrage is a methodical way to trade the differences in carrying costs between crypto instruments. The aim is to collect funding while offsetting much of the underlying asset’s directional price exposure. The work lies in finding a durable edge, building the hedge correctly and keeping both sides operational.

This guide is for readers who already understand spot trading, crypto wallets and the basics of exchange accounts. We will move from perpetual-futures mechanics to funding, hedging instruments, capital efficiency, costs, execution, risk management and API trading.

The attraction is easy to understand: you do not have to predict whether Bitcoin rises or falls to find a funding opportunity. But a position that starts approximately market-neutral can still lose money through changing funding, widening price spreads, borrowing costs, slippage, liquidation or a venue failure. Funding arbitrage requires active management; it does not deliver a guaranteed or passive income stream.

My focus throughout is practical: compare like-for-like rates, match the actual exposure, account for every cash flow and protect the capital needed to keep the hedge intact. A smaller executable edge can be more useful than a spectacular annualised number that disappears before both orders fill.

2. Understanding perpetual futures

Definition and history

A perpetual future is a derivative with no scheduled expiration. It provides long or short exposure to an underlying asset, usually through cash-settled profit and loss rather than a promise to take delivery on a future date. Positions can remain open while the product continues to exist and the trader meets its margin requirements. [3]

Robert Shiller’s proposal for perpetual futures appeared in a Cowles Foundation discussion paper in 1992 and in The Journal of Finance in 1993. BitMEX announced its XBTUSD perpetual leveraged swap on 12 May 2016 for launch on 13 May. These are distinct milestones: an academic proposal and a crypto-market implementation. [1] [2]

Market size and the rise of onchain trading

Perpetuals are a substantial part of crypto trading. CoinGecko reports that its top-ten centralised perpetual-exchange sample traded approximately $28.0 trillion in 2023, $58.5 trillion in 2024 and $86.2 trillion in 2025. Its top-ten decentralised perpetual-exchange sample recorded $647.6 billion in 2023, approximately $1.5 trillion in 2024 and $6.7 trillion in 2025. These figures describe those samples, not a complete census of worldwide derivatives activity. [19] [4]

2026 update. CoinGecko’s Q2 report puts top-ten centralised perpetual-exchange volume at $12.7 trillion for April–June 2026, down 10.0% from $14.1 trillion in January–March. These are quarterly totals, not full-year figures. [17]

For onchain activity, CoinGecko’s separate perpetuals study reports average monthly volume of $611.57 billion across its top-twelve decentralised perpetual exchanges in January–April 2026, compared with $531.65 billion during 2025. Decentralised venues accounted for 13.5% of combined perpetual open interest in that study as of 30 April 2026. Open interest adds a view of outstanding positions alongside trading turnover; neither measure establishes executable order-book depth. [18]

The annual table combines 2023–2024 figures from State of Crypto Perpetuals 2024 with 2025 figures from the 2025 Annual Crypto Industry Report. Both use top-ten samples, but membership can change over time; this is not a fixed panel of identical exchanges. B = billion; T = trillion. The separate 2026 perpetuals study covers eleven centralised and twelve decentralised exchanges and reports different 2025 totals. Its monthly averages should not be spliced into the annual series as a like-for-like revision.

Onchain venues expand the available combinations for CEX–DEX and DEX–DEX hedges. They also differ substantially in order-book design, pool architecture, collateral, settlement, fees and control over deposited assets. Assess each venue individually: “DEX” does not by itself guarantee unrestricted access, deep liquidity, low costs or freedom from custody and smart-contract risks.

How the contract works

Position and margin. A long gains when its contract price rises; a short gains when it falls. For a simple linear contract, price P&L is the underlying quantity multiplied by the price change, with the sign reversed for a short. Inverse contracts use different payout mathematics. Read the contract specification before combining them.

Leverage. Posting $1,000 to control $10,000 of notional exposure is 10× leverage on that margin. A 1% adverse price move is approximately $100 of loss before costs, or 10% of the initial margin. Liquidation can occur before the entire margin is exhausted because maintenance margin and other charges still apply.

Index, mark and last price. The index generally represents the underlying reference market. The mark is a venue-defined valuation used for functions such as unrealised P&L and liquidation. The last traded price is an execution observation. These prices may differ, and the mark is not simply another name for the spot index. [5] [12]

Funding. Payments create an economic incentive for the perpetual to remain near its reference market. This is an incentive, not a guarantee of convergence. Funding debits can themselves reduce the margin available to maintain a position.

Neither contract type is universally cheaper, simpler or safer. Leverage, liquidity and costs are product- and account-specific.

3. Funding rates explained

Who pays whom

Under the usual convention, positive funding means longs pay shorts; negative funding means shorts pay longs. If the relevant notional is $10,000 and the settled funding rate is +0.01%, the long pays $1 and the short receives $1 for that interval. Use the venue’s definition of funding notional and settlement currency. [5] [6]

Funding is often interpreted as a sentiment signal, but “more longs than shorts” is an incomplete explanation. Every matched futures contract has an opposing long and short. Aggressive demand for leverage can push the perpetual to a premium or discount; this price pressure then enters the funding formula. Some pool-based protocols explicitly incorporate long/short open-interest imbalance. [8]

How rates are calculated

There is no universal equation. Common ingredients include a premium index, an interest component, averaging, a clamp, a cap and a floor. The relevant premium can depend on executable impact bid and ask prices rather than a simple difference between the last price and the mark. Protocols may use different inputs or adaptive mechanisms. [5] [6] [8]

Separate direct inputs—premium, interest terms, clamps and protocol-specific imbalance—from indirect forces such as leverage demand, liquidity and volatility. A high-volume day is not automatically a high-funding day, and no unsupported bubble chart can assign universal importance to these factors.

Normalise the interval before comparing rates

If a genuine per-payment rate of 0.01% settles every eight hours, its simple annualisation is 10.95%. The same rate paid every hour annualises to 87.60%. However, an eight-hour-equivalent rate of 0.01% divided into eight hourly payments is 0.00125% per hour, and still annualises to 10.95%.

Hyperliquid documents hourly payments of one-eighth of an eight-hour-equivalent calculation. Its payment frequency does not create seven extra hours of free carry. On any venue, distinguish a displayed equivalent rate, the actual settlement interval, a forecast and an already settled rate. [7]

Sources: venue documentation [5] [6] [7] [9] [10]. Settings can change; this table is not live contract metadata.

Interpreting a large APR

Annualisation is an extrapolation, not a forecast or an APY. It does not assume realistic reinvestment or guarantee the rate will persist. The supplied design reference displays a 455.38% gross annualised spread; that number is a snapshot in a screenshot, not a verified trade recommendation or a promised return on total capital.

A universal Binance cap of ±0.75% per eight hours would be misleading. Read current caps, floors and intervals for the actual instrument. Check position eligibility at settlement rather than assuming every venue accrues funding identically or that an order submitted just before a timestamp will qualify.

4. Arbitrage strategies and hedging instruments

The basic trade

For comparable contracts, go long the lower-funding leg and short the higher-funding leg. Opposite signs are useful but not required: paying 0.01% on the long while receiving 0.03% on the short can create a positive gross difference. Compare the expected cash flows over the same holding window.

With $100,000 notional on each leg, short funding of +0.01% per eight hours and long funding of −0.005% per eight hours produce $10 plus $5 of funding receipts per interval. At unchanged notionals and rates, three intervals yield $45 gross per day. Price/basis P&L and all costs remain separate.

Spot

A common positive-funding trade is long spot and short the corresponding perpetual. Fully paid, unpledged spot has no derivatives-style liquidation merely because its price falls. It also has no perpetual funding payment. The futures side still needs sufficient margin, and the paired trade retains basis and venue risk.

Buying the spot asset consumes capital. Lending it or pledging it as collateral introduces additional claims and restrictions: the asset may no longer be immediately available to close or rebalance the hedge. A spot short ordinarily requires borrowing the asset before selling it.

Perpetual futures

Two perpetuals make long and short exposure accessible without separately borrowing the underlying coin for the short. They also create two funding streams and potentially two liquidation engines. Fees are not universally lower than spot fees; compare your actual account schedules and likely fills.

Margin spot

For negative funding, a possible structure is long perpetual and short borrowed spot. The long may receive funding while the spot short offsets price exposure. Borrow availability, financing rates, collateral requirements and repayment quantities determine whether it is feasible.

Borrowing creates an interest liability. It does not ordinarily let you earn interest merely by borrowing an asset. Any yield earned on separate collateral is another cash flow with its own risks. Borrowing rates can change, capacity can disappear, and margin spot can be liquidated.

Match exposure, not the number on the order ticket

For two linear instruments on the same asset, matching underlying units is generally more precise than blindly matching displayed dollar values at different prices. Confirm contract multipliers, token bundles such as 1,000-unit contracts, settlement currency and quantity increments. An identical ticker or order quantity does not establish identical exposure.

Inverse contracts, collateral denominated in the underlying asset, wrapped tokens and differently constructed indices need additional analysis. A hedge that offsets one price risk may still retain another.

5. Maximising returns and capital efficiency

Leverage changes the denominator

Funding applies to position notional; investment return is measured against capital committed. Increasing notional relative to margin can raise return on posted margin without improving the economic funding rate. Reserve capital, spot purchases and collateral on the other venue belong in the denominator when evaluating the whole strategy.

Use decimal rates over the same interval. With $5,000 margin and 3× leverage on each leg, both notionals are $15,000. A short rate of +0.01% and long rate of −0.005% generate $1.50 + $0.75 = $2.25 per interval. Against the $10,000 of posted margin, that is 0.0225% per interval, or 0.0675% over three identical intervals, before costs and basis changes. Additional reserves reduce the return on total committed capital.

No fixed leverage level is safe for every instrument. Size positions from stressed margin needs, order-book depth and the time required to repair a hedge. A gain on one exchange cannot automatically cover a loss on another. Research on crypto carry identifies margin stress and liquidation constraints as reasons apparent arbitrage can persist. [13]

Lending spot and using collateral

Lending a spot hedge may add income, but check withdrawal delays, redemption capacity, counterparty exposure and whether the asset can still serve its intended purpose. Do not count the same collateral twice or assume it can be freely lent, pledged and withdrawn simultaneously.

If an exchange recognises spot as collateral, understand its haircut and how collateral value changes in a market shock. Portfolio offsets are platform-specific. Yield earned on collateral or paid in a platform token should be modelled separately from funding and valued conservatively. Additional yield can come with restrictions that impair the hedge precisely when flexibility matters most.

Accessing unrealised P&L

Some products permit settlement or withdrawal of profit while positions remain open; others restrict this through margin or settlement rules. Confirm what is actually withdrawable, any fees and what the withdrawal does to the remaining position’s margin. Unrealised gains should not be treated as cash already available on another venue.

Price spread tactics

Buying the cheaper leg and selling the more expensive leg can add a gain if their spread narrows. For equal underlying quantity q, define the basis as the short-leg price minus the long-leg price. The paired price P&L is q × (entry basis − exit basis), before costs. A widening basis produces a loss.

A spread may reflect differing indices, local order flow, withdrawal restrictions or venue credit risk. Convergence is not assured. Evaluate funding and entry prices together; a favourable funding leg can trade at a price that offsets its apparent advantage.

Holding periods

Stay in a trade while its expected remaining net carry justifies its risks and exit costs. Longer holding can spread entry and exit fees over more funding events, but it also extends exposure to rate reversals and venue problems. Constant switching for a slightly higher rate can consume the entire edge. There is no universal seven-day maximum or funding threshold that determines an optimal exit.

PEPE case study from March 2024

The original example paired a Binance PEPE perpetual with OKX margin spot and described $10,000 of capital. Its narrative reported $508 profit, 5.08% monthly return and $16.39 average daily profit; its embedded summary reported $538, 5.38% and $17.35. The research audit found no reproducible primary dataset resolving those differences.

Neither figure can be presented as a validated historical return. Claims of 30 profitable days out of 31 or “steady, low-risk returns” likewise need evidence beyond an unverified chart. The example remains useful as an accounting exercise.

To rebuild that back-test, collect timestamped realised funding, the relevant borrow rates and accrual rules, borrow availability, exact quantities, contract multipliers, entry and exit fills, the fee tier, slippage assumptions and historical specification changes. Align timestamps and currencies, account for both legs and include all committed capital. A graph of rate differences alone is not a net-return back-test.

6. Understanding fees and costs

A funding spread is gross income. The result that matters is what remains after operating the whole hedge. Include both entry trades, both exit trades and any rebalancing, as well as carrying costs during the holding period.

For accounting based on actual fills, execution slippage is already reflected in realised price P&L. Subtract it separately only when the price/basis calculation uses benchmark prices; otherwise you count the same cost twice.

Trading fees

Maker and taker fees can differ by product, rolling volume, account tier and location. A limit order that crosses the book may execute as a taker; it is not a maker order merely because it has a limit price. Passive orders may reduce explicit fees but bring non-fill and adverse-selection risk. Compare executable net returns instead of ranking exchanges by an old fee table. [11]

If expected net funding before those costs is $3 per day, the simple recovery period is ten days, ignoring basis changes and borrowing. If borrowing costs $1 per day, it becomes fifteen days. These are invented assumptions to demonstrate arithmetic, not current exchange fees or a forecast. The edge can vanish before the recovery period ends.

Borrowing, gas and transfers

Borrow interest may accrue in discrete units and may be paid in the borrowed asset. Include minimum accrual periods, changing rates and the eventual repayment quantity. Borrow capacity observed during research may be unavailable when the order is placed.

For onchain trading, identify which actions actually incur fees: deposits, orders, settlement, withdrawals, bridging or keeper execution. Different architectures charge differently; do not assume every DEX order incurs an Ethereum-style gas transaction. A cheaper route may introduce bridge, liquidity or timing risk.

Also account for capital left idle as a reserve, stablecoin conversion spreads and the operational burden of maintaining another venue. Incentives and token rewards are separate uncertain income, not a substitute for positive trading economics.

7. The trading process

When building a shortlist, ArbiScan is one tool that can help you compare funding spreads, inspect both legs and review rate history. Use it as a starting point for research, then check live exchange data and the full cost of the trade before acting.

Opening positions

1. Identify a comparable pair. Confirm the underlying asset, contract type, quote and collateral currencies, multiplier, quantity precision, position limits and current trading status.
2. Validate the opportunity. Normalise the rates and distinguish forecasts from settled observations. Review enough history to see reversals and stressed periods. Estimate cash flows over a specific holding window.
3. Check executable liquidity. Inspect both order books at the intended size and stress the exit. A multiple of 24-hour volume is not a substitute for available depth.
4. Budget the whole trade. Include fees, adverse fills, funding on both legs, borrowing, transfers and plausible basis losses. Specify when the trade should be paused, repaired or closed.
5. Prepare collateral and borrow. Confirm sufficient usable margin on each venue, actual borrow availability and open-interest capacity before creating a dependent leg. Keep reserves where they can be used in time.
6. Execute and reconcile fills. Match the underlying exposure. Track partial fills, cancelled orders and remaining quantities. A pair of submitted orders is not yet a hedge.

A marketable limit order can cap the worst acceptable execution price, but may still fill only partly or not at all. A passive order may wait while the market moves. Decide in advance how long unmatched exposure may remain and what repair or unwind action follows a rejected or unfilled leg.

Ongoing monitoring

Monitor net underlying exposure, each account’s maintenance margin and equity, current mark/index dislocations, funding forecasts, realised funding receipts, borrowing rates and remaining liquidity. Watch exchange status, withdrawal availability, collateral haircuts and changes to the contract itself.

Use alerts based on the position’s tolerances and the time needed to respond. A 2% price move, a 20% distance from liquidation or a 0.005% funding change is not a universal safety boundary. Record the trade’s combined liquidation-value P&L, including what it would cost to close now.

Closing positions

Reassess the expected remaining edge against exit costs and risks. Exit or reduce when the economics no longer work, the hedge cannot be maintained, margin becomes inadequate or venue conditions change. Do not keep a deteriorating trade solely to recover costs already paid.

Coordinate both closing legs and use reduce-only instructions where supported and appropriate. Confirm fills and cancel residual orders; an old working order can reopen exposure after the hedge is gone. Repay any spot borrowing, check accrued interest and verify that neither account retains an unintended position or liability.

Finally, reconcile funding cash flows, price P&L, fees, transfers and changes in account equity. Preserve timestamps, quantities and source records. Compare the realised result with the entry model and investigate differences before scaling the strategy.

8. Risk management

Market, basis and liquidation risk

A delta hedge reduces sensitivity to a shared underlying price move. It does not eliminate basis changes or local liquidation. A losing short on one exchange needs margin there, even if an offsetting spot or long position is gaining elsewhere. Transfer delays and withdrawal suspensions make this separation especially important. [13]

Size for adverse price moves, widening basis, collateral depreciation and the costs of an urgent exit. The relevant liquidation rules depend on mark price, maintenance tiers, margin mode and other positions sharing the account. Adding margin can buy time but increases capital exposed to that venue; it is not automatically the correct response. [12]

Liquidity and borrowing risk

Order-book liquidity can disappear during stress. Test expected fills at realistic sizes, allow for wider spreads and avoid sizing solely from headline volume or open interest. Limit orders constrain prices but do not guarantee an exit. Borrowing limits, recalls or rate changes can undermine a margin-spot hedge even when funding remains attractive.

Venue, collateral and protocol risk

Centralised venues introduce custody, solvency, account-access and withdrawal risks. Onchain venues introduce contract, oracle, validator or sequencer, governance and bridge risks depending on their design. Deposits in two venues are two separate exposures; diversification can reduce concentration while adding operational complexity.

Stablecoins and wrapped assets are not interchangeable cash. Depegs, conversion restrictions and collateral haircuts can change both the hedge and its ability to survive. A liquid market on one venue does not make collateral transferable from the other.

Technology risk

An extreme funding observation is a reason to investigate, not proof of a technical glitch or a profitable opportunity. Check timestamp freshness, rate units, contract status, index and mark inputs, official notices and actual executable prices. The original draft’s explanation of unexplained spikes as exchange resets was not supported by its chart.

Venue protections can change during a crisis. Binance’s June 2026 notice applying a temporary Last Price Protected mechanism to HUSDT after a security incident and spot-market disruption illustrates why current product notices matter. It does not establish the cause of an unrelated funding spike. [14]

Operational and security risk

Common failures include confusing contracts with underlying units, mixing hourly rates with annualised rates, duplicate orders after a timeout, stale market data and closing the wrong position side. Keep a checklist and verify actual account state after every ambiguous execution response.

Use official access points, protect seed phrases and restrict trading keys to necessary permissions. Keep withdrawal capability separate from routine trading automation where possible. Evaluate wallet approvals and third-party applications before using them; a platform ranking does not prove safety.

Access, regulation and records

Eligibility for derivatives, account restrictions and tax treatment depend on the trader, venue and jurisdiction. Using an overseas venue does not by itself resolve those obligations. Keep complete trade and transfer records and obtain jurisdiction-specific advice where necessary. This guide does not determine legal eligibility or a personal tax treatment.

9. Advanced techniques and API trading

APIs let software collect market data, manage orders and reconcile accounts. REST or HTTP interfaces are useful for requests and snapshots; WebSockets support streaming updates. Automation can improve consistency and response time, but also scale a wrong assumption or sizing bug faster than manual trading.

Current exchange documentation defines connection behaviour, message limits, authentication and available order interfaces. Treat these as live implementation requirements. Binance publishes WebSocket requirements, and dYdX provides integration documentation for programmatic access. [15] [16]

A practical architecture

1. Market-data normalisation. Convert symbols, quantities, multipliers, rate periods and timestamps to a consistent representation. Preserve the original values for diagnosis. Reject stale or incomplete input.
2. Opportunity engine. Estimate funding over a common horizon, executable basis and the full cost budget. Check borrow capacity, exposure limits and reserves before producing an order intent.
3. Paired-execution state machine. Track orders from intent through acknowledgement, partial fill, completed hedge, repair and exit. Make unmatched quantities explicit.
4. Independent risk controls. Limit per-venue exposure, net delta, margin stress and unhedged time. Halt new entries on abnormal data or venue states while retaining a controlled path to manage existing positions.
5. Audit and reconciliation. Store order identifiers, fills, funding, fees and balances. Regularly compare internal state with the exchange’s actual positions and liabilities.

Handling failure

A timeout does not prove an order failed. Before retrying, query the original order and account state using identifiers supported by the venue. Apply rate-limit backoff, reconnect streams deliberately and rebuild a verified snapshot when updates are missing. Restarting the bot should not create a second copy of a live hedge.

A kill switch needs a defined effect: stopping new entries is different from cancelling orders or unwinding positions. Uncoordinated “close everything” actions can increase slippage or strand a single leg. Test failure paths, including one venue offline, one partial fill, lost connectivity and a rejected reduce-only order.

Testing and deployment

Begin with recorded-data replay and paper trading, then testnets where available. Testnets help validate logic but may not reproduce live liquidity, borrowing, funding or liquidation. Use carefully limited live sizing only after position accounting and failure recovery are verified. Keep human monitoring, alerts and an accessible recovery procedure.

Use minimal key permissions, protect secrets, respect venue limits and retain complete logs without exposing credentials. Automated trading remains subject to applicable exchange rules and record-keeping obligations.

10. Key takeaways and best practices

Funding arbitrage works only when the whole position works. A rate difference is a starting point; the executable hedge, capital distribution and after-cost outcome determine the result.

• Understand the contract. Confirm the underlying units, payout, margin and settlement rules.
• Normalise funding. Compare actual per-period cash flows over one common horizon.
• Match the hedge. Reconcile fills and underlying exposure, including multipliers and collateral risks.
• Use the right return denominator. Include capital committed to both legs and operational reserves.
• Price the full round trip. Funding, basis, fees, slippage, borrowing and transfers all affect the result.
• Protect local margin. Profits trapped elsewhere cannot automatically prevent liquidation.
• Plan repairs and exits. Treat the two legs as one trade when a stop, rejection or outage occurs.
• Validate performance claims. Historical charts require reproducible data and all-in accounting.
• Automate cautiously. Monitor data quality, order state and actual exchange balances.
• Stay adaptable. Recheck mutable venue settings and learn from reconciled outcomes.

Conclusion

Funding rate arbitrage offers a systematic way to explore crypto-market inefficiencies with reduced directional exposure. It can still produce significant losses. The durable skills are careful measurement, disciplined execution and the ability to keep a hedge intact under stress.

Start with a trade you can explain cash flow by cash flow. Know why the rate differs, what could erase the edge, where the margin sits and how both legs will close. That understanding is more valuable than any headline APR.

Appendix A. Working glossary

Perpetual future

A derivative with no scheduled expiry, generally using funding to help align its price with a reference market.

Funding rate

The signed rate used to calculate a funding payment under a venue’s rules.

Notional

The economic size of a position, distinct from the collateral posted to support it.

Delta

Sensitivity to a change in the underlying price. A delta hedge offsets this sensitivity approximately.

Basis

The difference between the derivative and spot price, or between the two prices in a paired trade; state the sign convention.

Index price

The reference price or basket used by a venue for an underlying market.

Mark price

A venue-defined valuation used for risk and accounting. It can differ from both the index and last trade.

Initial and maintenance margin

Collateral requirements for opening and maintaining a leveraged position.

Liquidation

Forced reduction or closure when margin requirements are breached. It is not a promise that all losses will be limited to the original margin.

Open interest

Outstanding derivative exposure, reported in a specified unit. It is not the same as executable liquidity.

Maker and taker

Orders or fills that add resting liquidity and those that remove available liquidity, respectively.

Slippage

The difference between a reference execution price and the actual fill.

Borrow rate

The financing charge for borrowed funds or assets, with an associated accrual convention.

Realised and unrealised P&L

Profit or loss recognised through settlement or closure versus valuation changes on open positions; treatment varies by product.

APR and APY

Simple annualised rate versus an annual yield incorporating specified compounding assumptions. Neither guarantees future returns.

CEX and DEX

Centralised and decentralised exchange categories. Actual custody, execution and settlement arrangements vary within them.

Oracle

A mechanism supplying external reference data to a protocol.

Bridge

A system moving or representing assets across networks, introducing its own costs and risks.

Appendix B. Substantive research corrections

This edition preserves the manuscript’s progression and central idea while correcting statements that materially affect the interpretation of the strategy.

Sources

Exchange parameters are mutable; consult the applicable live contract and account documentation before relying on them. Numerical trading examples are illustrative unless explicitly identified as historical market data.

1. Robert J. Shiller — Measuring Asset Values for Cash Settlement in Derivative Markets

   Cowles discussion paper, 1992; journal publication in 1993.

2. BitMEX — Announcing the launch of the perpetual XBTUSD leveraged swap

   Announcement of the May 2016 launch.

3. BitMEX — What is a perpetual contract

   Contract definition and settlement.

4. CoinGecko — 2025 Annual Crypto Industry Report

   2024–2025 volumes for the top-ten perpetual CEX and DEX samples.

5. Binance — Introduction to Binance Futures funding rates

   Funding conventions, timing and contract-dependent parameters.

6. OKX — Perpetual funding fee mechanism

   Formula, interval scaling, payments and schedule adjustments.

7. Hyperliquid — Funding

   Hourly payments and eight-hour-equivalent funding calculation.

8. GMX — Fees

   Protocol-specific funding and borrowing design.

9. dYdX — Default funding rates

   Hourly funding and market parameters.

10. Bybit — Funding fee calculation

    Payment calculation and contract-specific funding intervals.

11. Hyperliquid — Fees

    Volume-tiered fee schedules.

12. Binance — Futures liquidation protocols

    Mark-price and liquidation mechanics.

13. Bank for International Settlements — Crypto carry

    Working paper on carry and limits to arbitrage.

14. Binance — HUSDT Last Price Protected mechanism notice

    15 June 2026 notice; an example of changed protections during market disruption.

15. Binance developer documentation — WebSocket market streams

    Connection behaviour and implementation requirements.

16. dYdX — Integration documentation

    Programmatic integration interfaces.

17. CoinGecko — 2026 Q2 Crypto Industry Report

    Top-ten centralised perpetual-exchange volumes for Q1 and Q2 2026.

18. CoinGecko — State of Crypto Perpetuals Report 2026

    January–April 2026 activity across eleven centralised and twelve decentralised perpetual exchanges; open interest as of 30 April 2026. Published study updated 21 May 2026.

19. CoinGecko — State of Crypto Perpetuals 2024

    2023 and 2024 annual trading volumes for top-ten centralised and decentralised perpetual-exchange samples.