Funding cross-exchange arbitrage:
delta-neutral income from the rate spread
The funding rate is the mechanism perpetual futures use to keep their price near spot. The rate changes constantly and doesn't match across exchanges. Long where the rate is positive (paid to long-holders), short where it's negative (paid to short-holders) — you collect the difference, direction doesn't matter.Guide: how funding payments work, how to compute divergence across exchanges with different cadences, how to defend against liquidation risk and rate reversals.
Funding-rate arbitrage in plain English
A perpetual future is a contract with no expiration date. To keep its price near spot, exchanges use the funding rate: a periodic payment from one side of the contract to the other. When long positions outnumber shorts — longs pay shorts, and vice versa. The exchange sets the payment size and cadence.
Where the profit comes from
The same coin (BTC, ETH, SOL) trades as a perpetual simultaneously on dozens of exchanges. Funding rates on these exchanges almost never match: +0.01% per 8h on one, +0.06% on another, negative on a third. That means — open a long on the exchange with a high positive rate (collect payments) and simultaneously a short on the exchange with a low or negative rate (pay less or even get paid). Equal-sized positions = delta-neutral (price moves — both sides compensate). Profit comes from the funding-rate difference.
Spot cross-exchange: buy on spot → move the asset through a blockchain → sell on the spot of another exchange. Profit from the price difference, hold minutes.
Futures cross-exchange: long perp on the cheap exchange + short on the rich one simultaneously. Profit from the price gap closing, hold hours.
Funding cross-exchange (this guide): same long+short structure on two perp exchanges, but the focus is on the funding-rate difference, not the price gap. Profit accumulates with each funding payment, hold runs days to weeks. The slowest and steadiest type: less timing stress, even APR.
Adjacent strategies without direct perp-perp:
CEX-DEX — exchange vs on-chain pool. Spot-Futures — spot vs perp of the same asset on one exchange (basis trade), not strictly cross-exchange.
Why rates diverge between exchanges
On an exchange with a heavy skew in open interest one way (e.g. many longs), the funding rate is positive: that's how the exchange keeps the contract price from running too high. On another exchange with the opposite skew — the rate is negative. These skews arise from different user bases: Hyperliquid attracts high-leverage degens, Binance more conservative ones, so the rates differ.
Where persistent divergences usually show up
How to compute profit and where the costs sit
Funding arbitrage is the 'calmest' form of arbitrage, but it requires understanding the cadence math, rate normalisation, and where fees accumulate.
Cadence math: normalising to 8 hours
Exchanges pay funding with different frequencies. To compare rates directly, you have to normalise them to a single window — usually 8 hours:
- Hyperliquid (1h cadence): 0.005% × 8 = 0.04% per 8h
- Bybit (8h cadence): 0.01% per 8h
- EdgeX (4h cadence): 0.005% × 2 = 0.01% per 8h
In this example: open long on Hyperliquid (+0.04% per 8h) and open short on Bybit (+0.01% per 8h). Net: +0.03% per 8h = +0.09% per day = ~2.7% per month on pure funding spread. On a $10,000 position — $90 per week, $360 per month at stable rates.
Costs on open + close
Each position pays a taker fee on opening and closing:
- Open long on exchange A: 0.04% of volume
- Open short on exchange B: 0.04% of volume
- Close both sides: another 0.08%
Round-trip total ~0.16% on entry + exit. To recover those fees from net funding income, the divergence has to be at least 0.04% per 8h on a hold ≥ 4 cycles (32 hours). Less than that — the fees eat it.
How to open and close a funding-arbitrage position
The algorithm is slower than other types — minutes to open, then hours or days to hold. So the emphasis is on solid prep and an exit plan, not reaction speed.
- Prep
-
01
Find a divergence of ≥ 0.04% per 8h
Through a scanner (Finder pings Telegram with net projected over 1/3/7 cycles) or manually comparing current funding rates across exchanges. Ignore divergences below break-even (~0.04% per 8h on typical 0.04% taker fees).
-
02
Assess rate stability
Open each side's funding-rate history for the past 24–72 hours. If the rate has been stable or rising since onset — the situation is stable. If it's already close to a reversal (historical flip pattern after N hours) — too risky. The 'Next flip' column in the dashboard helps assess.
-
03
Prepare margin on both exchanges
Both perp exchanges need USDT/USDC margin. Equal-sized positions = delta-neutral. Leverage 3×–5× keeps liquidation risk low. Avoid high leverage even when it looks 'safe' — funding arb wins through patience, not through timing.
- Execute
-
04
Open long on the exchange with the positive rate
Market order on the side that collects funding. Taker fee (~0.04%) is factored into the payback calculation. Record exact price and time — for monitoring later.
-
05
Open short on the exchange with the low/negative rate
Right after opening the long — short on the opposite side at the same size. The position is now delta-neutral: price can move — both sides compensate. Direction PnL is zero, only funding payments accumulate.
- Hold
-
06
Collect funding through cycles
Every 1/4/8 hours (depending on the exchange's cadence) a funding payment lands on one side and is debited from the other. Net = the spread per cycle. On 0.04% per 8h divergence, a $10k position collects $4 per cycle, $12 per day, ~$360 per month at stable rates.
-
07
Monitor rates for reversal
Once a day (or more often in volatile markets) check current rates on both sides. If divergence has shrunk below 0.02% per 8h — the position no longer covers its close-out fees. If the rate on the collecting side has flipped negative — the position is losing, time to close.
- Close
-
08
Close both sides simultaneously
Close with market orders on both exchanges simultaneously. To prevent price from moving against one side before the other closes — close within 30 seconds. Net profit = (accumulated funding) − (open + close fees). Record in the log: average divergence, hold length, total PnL.
Five risks of funding arbitrage
Funding-rate reversal
The rate on the side you were collecting from can reverse (was +0.05%, becomes −0.03%). From that point the position loses with every cycle. Mitigation: daily monitoring + a flip predictor based on 24h history. A good exit plan: close when divergence has shrunk to < 50% of original.
Liquidation on one side
During a volatility spike, price can move far enough to trigger a margin call on one side (even with neutral net direction). If liquidation happens — the position is no longer delta-neutral, the remaining side carries directional risk. Mitigation: leverage ≤5× and cross-margin (if supported) — letting one side's PnL offset the other's margin requirements.
Exchange technical issues at close time
One of the exchanges can temporarily go down (degraded mode, upgrade, API incident). If the position needs to close but the exchange is unavailable — the remaining side carries directional risk. Mitigation: use proven exchanges with aggressive uptime SLAs for funding arb, not fresh exchanges.
Divergence too small after fees
Signal shows 0.04% per 8h divergence, but during open the price moved a bit — actual divergence at entry is already 0.025%. Round-trip fees of 0.16% won't be recovered even over a week of hold. Mitigation: check the current divergence at entry, not the signal value.
Can't short on one side
On some exchanges, for certain tokens, shorts can be paused (long-only) — e.g. after a forced delisting. If you were planning to short there — you won't be able to open it. Mitigation: check short availability on the contract page before opening the long side.
BTC long Hyperliquid (1h cadence) + short Bybit (8h), 7-day hold
Step by step — a realistic funding-arbitrage trade. Numbers model a typical divergence under normal market conditions: Hyperliquid pays +0.005% per 1h (= +0.04% per 8h normalised), Bybit pays +0.01% per 8h. Net divergence 0.03% per 8h. Size $10,000 on each side.
That's ~22% annualised on capital at stable rates. Seems modest, but the upside: (a) risk is bounded by one-sided liquidation, (b) zero directional risk, (c) the position scales to the limits of the exchanges' margin requirements.
On a $100,000 position ($100k per side) the same math gives ~$430 per week or $20,000 per year at preserved rates. That's comparable to tradfi income strategies, but on a crypto risk scale.
If the rate had reversed 3 days into the trade (Hyperliquid −0.005% per 1h), the remaining 4 days would have been losing. Net result would have been ~$0 after fees. So daily monitoring and a solid exit are mandatory.
Funding-arbitrage checklist
- Funding-rate divergence ≥ 0.04% per 8h after cadence normalisation.
- Rate history for the past 24–72h on both sides checked — no recent reversals.
- Margin on both exchanges sufficient for the position at leverage ≤5×.
- Cross-margin enabled (where supported) — resilience to a one-sided margin call.
- Contract type matches on both exchanges (USDT-perp ≠ coin-margined ≠ inverse).
- Short is allowed on the side where you plan it (some exchanges temporarily disable for certain tokens).
- Exit plan ready: close when divergence shrinks to < 50% of original or when the rate reverses.
- Position sizes equal = delta-neutral in fact, not approximately.
Common funding-arb mistakes
What you can and can't expect
Funding arbitrage is the slowest and steadiest type of arbitrage in crypto. It's not day trading — it's holding delta-neutral positions for days and weeks. Income drips in with each funding payment, no adrenaline and no timing stress.
What drives results
Main factor — divergence persistence between exchanges. In stable markets rates can diverge for weeks; in volatile events rates can reverse in hours. The same signal will deliver different results to two traders depending on how fast they react to a reversal.
What's math, not luck
On 0.04% per 8h divergence and 0.16% round-trip fees, break-even is around 32 hours of hold. Holding longer = profit. Less = loss. That's a hard threshold, not an assumption.
- Typical annual return on funding arbitrage: 15–35% APR on capital at moderate leverage. During peak events (memecoin seasons) — up to 100%+ APR for short periods.
- Income scales linearly with position size. $1k = ~$30/month. $100k = ~$3000/month.
- Delta-neutral position = zero directional risk. Price moves — both sides compensate. The only risk is liquidation on one side during a volatility spike.
How funding arb differs from other types
Unlike cross-exchange Spot-Spot or CEX-DEX, there's no asset transfer between exchanges — both positions are open on different exchanges at the same time. Unlike Futures-Futures, where closing happens in hours when prices converge, here the position is held longer — as long as the rates stay favourable.
Built for exactly this type of arbitrage
Funding arbitrage requires constant monitoring of dozens of perp exchanges with different cadences — you can't keep the whole matrix in your head manually. Finder does it in the background: subscribes to the funding streams of every supported exchange, normalises rates to an 8h window, computes divergence per contract, and pings Telegram only when divergence × expected hold horizon delivers clean profit after fees.
What lands in your channel and what the dashboard looks like — below.
Every exchange with a funding stream
Different cadences are an opportunity, not a problem — the divergence between an hourly-funded exchange and an 8h-funded exchange is exactly what creates the edge.
What the funding-rate channel sends
Header carries the contract, the rate divergence per 8h, magnitude tier, and projected net at $1k for the recommended hold. Long and Short rows show funding rates inline so you immediately see who pays and who collects.
The FUNDING table sorts exchanges by rate. Top is the most positive (collect), bottom is the most negative (pay) — the recommended trade brackets the gap. Cadence column shows 1h / 4h / 8h so you know how often you'll receive or pay.
- Magnitude tiers: HIGH ≥0.08%/8h, MEDIUM ≥0.04%/8h
- Cycle countdown to the next funding payment
- Flip-prediction flag when historical rate suggests inversion likely
- Projected net for 1 / 3 / 7 funding cycles inline
SOL +0.092% /8h ⚡(9$ /8h on $1000) | PERP
1 cycle (8h): +$9 · 3 cycles (24h): +$27 · 7 cycles: +$63
⚠ HL rate likely to flip within 4–6h
Funding cycles synced, flips projected
All funding pairs firehose — every contract, every exchange cross sorted by funding-rate divergence. The cycle column tells you when the next funding payment lands so you size the hold to capture full cycles.
Next flips sorts by predicted rate-flip time. Useful for closing positions before the rate inverts and you start paying instead of collecting.
- 1h / 4h / 8h cycle annotated on every row
- Net per 8h normalised so cross-cadence rates are comparable
- Flip-prediction based on 24h rate history
- Click a row → side-by-side funding history charts
Funding cross-exchange is free during beta
All arbitrage types are included with no tiers or quotas while the product is in beta.
Full feature access while we polish
All arbitrage types, every exchange, every alert tier — no plans, no card, no quota. We'll announce pricing before launch; early users keep grandfathered terms.
- Every CEX–CEX, CEX–DEX, DEX–DEX and funding alert
- Telegram bot + web dashboard, both included
- All 25+ exchanges, all networks, full asset coverage
- Pre-flight: open routes, taker fees, fillable depth checked on every signal
Practical questions about funding-rate arb
How is this different from Futures–Futures spread arbitrage?
What does "+0.05% per 8h typical" actually pay at scale?
Funding rates flip — how do I avoid the wrong side?
Hyperliquid funds hourly — does that change the math?
What's the smallest divergence still worth it?
Different shape, different rhythm.
Delta-neutral income from the funding-rate spread
So you don't have to track rates across dozens of perp exchanges by hand — the ping arrives when divergence is normalised to an 8h window and the hold covers round-trip fees. All the math is in the message.
Get alerts