Self-collected data
Cross-venue funding divergence
The gross funding spread between comparable BTC and ETH linear perpetuals, from one 15-minute sweep. Same rate type, same normalization, same timestamp.
26 BTC venuesLive data
19.6%
widest BTC spread, annualized · dYdX vs BloFin
30.9%
widest ETH spread, annualized · dYdX vs GMX
210
executable pairs in the sweep of 11:53 UTC
BTC perpetual funding divergence
BTC — widest executable venue pairs (12 of 105 executable pairs)
Sweep 09-09 11:52 UTC| Long leg | Short leg | Long rate 1h | Short rate 1h | Spread ann. | Min OI |
|---|---|---|---|---|---|
| dYdX | BloFin | 0.00003% | 0.00226% | 19.6% | $15.3m |
| dYdX | GMX | 0.00003% | 0.00217% | 18.8% | $15.3m |
| Phemex | BloFin | 0.00039% | 0.00226% | 16.4% | $194.0m |
| Phemex | GMX | 0.00039% | 0.00217% | 15.7% | $24.0m |
| Bybit | BloFin | 0.00065% | 0.00226% | 14.1% | $427.2m |
| Bybit | GMX | 0.00065% | 0.00217% | 13.4% | $24.0m |
| dYdX | Kraken | 0.00003% | 0.00143% | 12.3% | $15.3m |
| Phemex | Kraken | 0.00039% | 0.00143% | 9.1% | $165.3m |
| Bybit | Kraken | 0.00065% | 0.00143% | 6.8% | $165.3m |
| KuCoin | Hyperliquid | 0.00079% | 0.00125% | 4.1% | $1.11b |
| KuCoin | Orderly Network | 0.00079% | 0.00125% | 4.1% | $2.4m |
| KuCoin | GRVT | 0.00079% | 0.00125% | 4.1% | $199.6m |
Executable means that both legs reported open interest of $1,000,000 or more in this sweep. Rates are normalized to one hour. The spread is gross: it holds no trading fee, no slippage and no borrow cost. The rows keep the widest spread first, and one venue fills at most 3 rows, so the table shows different venue combinations instead of repeats of one leg.
BTC — venue matrix (15 venues with reported open interest)
Sweep 09-09 11:52 UTC| Venue | Type | Normalized 1h | Annualized | Open interest | Interval | Executable |
|---|---|---|---|---|---|---|
| BloFin | cex | 0.00226% | 19.8% | $427.2m | 8h* | Yes |
| GMX | dex | 0.00217% | 19.1% | $24.0m | 1h* | Yes |
| Kraken | cex | 0.00143% | 12.5% | $165.3m | 1h | Yes |
| Hyperliquid | dex | 0.00125% | 11.0% | $2.84b | 1h* | Yes |
| Orderly Network | dex | 0.00125% | 11.0% | $2.4m | 8h* | Yes |
| GRVT | dex | 0.00125% | 11.0% | $199.6m | 8h* | Yes |
| Crypto.com | cex | 0.00123% | 10.8% | $493.0m | 8h* | Yes |
| Lighter | dex | 0.00120% | 10.5% | $166.6m | 1h* | Yes |
| Paradex | dex | 0.00113% | 9.9% | $4.0m | 8h* | Yes |
| Bitget | cex | 0.00108% | 9.4% | $2.75b | 8h* | Yes |
| Deribit | cex | 0.00103% | 9.0% | $38.2m | 8h* | Yes |
| KuCoin | cex | 0.00079% | 6.9% | $1.11b | 8h* | Yes |
| Bybit | cex | 0.00065% | 5.7% | $4.44b | 8h | Yes |
| Phemex | cex | 0.00039% | 3.4% | $194.0m | 8h* | Yes |
| dYdX | dex | 0.00003% | 0.2% | $15.3m | 1h* | Yes |
8 more BTC venues published a current rate in this sweep without open interest, so this page cannot test them against the $1,000,000floor. An absent value stays absent and never becomes zero. “*” marks an interval taken from the venue standard schedule for majors.
ETH perpetual funding divergence
ETH — widest executable venue pairs (12 of 105 executable pairs)
Sweep 09-09 11:52 UTC| Long leg | Short leg | Long rate 1h | Short rate 1h | Spread ann. | Min OI |
|---|---|---|---|---|---|
| dYdX | GMX | -0.00160% | 0.00192% | 30.9% | $10.8m |
| dYdX | Kraken | -0.00160% | 0.00155% | 27.6% | $10.8m |
| dYdX | Hyperliquid | -0.00160% | 0.00125% | 25.0% | $10.8m |
| GRVT | GMX | -0.00006% | 0.00192% | 17.4% | $19.1m |
| Phemex | GMX | 0.00010% | 0.00192% | 16.0% | $19.1m |
| GRVT | Kraken | -0.00006% | 0.00155% | 14.2% | $79.3m |
| Phemex | Kraken | 0.00010% | 0.00155% | 12.8% | $79.3m |
| GRVT | Hyperliquid | -0.00006% | 0.00125% | 11.5% | $145.1m |
| Phemex | Hyperliquid | 0.00010% | 0.00125% | 10.1% | $245.3m |
| Bybit | Orderly Network | 0.00012% | 0.00125% | 9.9% | $25.9m |
| Bybit | BloFin | 0.00012% | 0.00125% | 9.9% | $292.2m |
| Bybit | Deribit | 0.00012% | 0.00122% | 9.6% | $24.8m |
Executable means that both legs reported open interest of $1,000,000 or more in this sweep. Rates are normalized to one hour. The spread is gross: it holds no trading fee, no slippage and no borrow cost. The rows keep the widest spread first, and one venue fills at most 3 rows, so the table shows different venue combinations instead of repeats of one leg.
ETH — venue matrix (15 venues with reported open interest)
Sweep 09-09 11:52 UTC| Venue | Type | Normalized 1h | Annualized | Open interest | Interval | Executable |
|---|---|---|---|---|---|---|
| GMX | dex | 0.00192% | 16.8% | $19.1m | 1h* | Yes |
| Kraken | cex | 0.00155% | 13.6% | $79.3m | 1h | Yes |
| Hyperliquid | dex | 0.00125% | 11.0% | $2.57b | 1h* | Yes |
| Orderly Network | dex | 0.00125% | 11.0% | $25.9m | 8h* | Yes |
| BloFin | cex | 0.00125% | 11.0% | $292.2m | 8h* | Yes |
| Deribit | cex | 0.00122% | 10.7% | $24.8m | 8h* | Yes |
| Lighter | dex | 0.00120% | 10.5% | $108.6m | 1h* | Yes |
| Paradex | dex | 0.00119% | 10.5% | $2.8m | 8h* | Yes |
| Bitget | cex | 0.00116% | 10.2% | $2.00b | 8h* | Yes |
| KuCoin | cex | 0.00065% | 5.7% | $473.8m | 8h* | Yes |
| Crypto.com | cex | 0.00033% | 2.9% | $435.7m | 8h* | Yes |
| Bybit | cex | 0.00012% | 1.1% | $1.89b | 8h | Yes |
| Phemex | cex | 0.00010% | 0.9% | $245.3m | 8h* | Yes |
| GRVT | dex | -0.00006% | -0.5% | $145.1m | 8h* | Yes |
| dYdX | dex | -0.00160% | -14.0% | $10.8m | 1h* | Yes |
9 more ETH venues published a current rate in this sweep without open interest, so this page cannot test them against the $1,000,000floor. An absent value stays absent and never becomes zero. “*” marks an interval taken from the venue standard schedule for majors.
How the spread is measured
A funding divergence is the gap between the funding rate of one venue and the rate of another venue for the same asset. PerpFinder reads the current rate of every covered venue in one sweep. It converts each rate to a common one-hour basis, then subtracts the smaller rate from the larger one. The result is a gross spread.
A trader who wants that spread holds two positions at the same time. The long leg sits on the venue with the lower rate. The short leg sits on the venue with the higher rate. The two positions cancel most of the price risk, and the funding difference remains. The sweep covers BTC and ETH, because those are the two assets that PerpFinder collects with full interval and settlement detail.
Each row of a pair table holds one long leg and one short leg from the same sweep. The spread column shows the gross difference between the two rates, extrapolated to one year at a constant rate. The minimum open-interest column shows the smaller of the two legs, because that value limits the size of the pair. A pair enters the table only when both legs reported open interest of $1,000,000 or more in that sweep.
Widest executable spread, last 30 stored days
preliminary- BTC: 30 stored days, 2747 sweeps · widest 152.0% on 2026-08-25 · typical 25.8%
- ETH: 30 stored days, 2748 sweeps · widest 162.5% on 2026-08-23 · typical 28.3%
Each point is the widest executable spread of one UTC day of 15-minute sweeps, extrapolated to one year. “Typical” is the median of the daily median spreads. The series starts at the first stored day inside the window. PerpFinder adds no synthetic backfill, and a day without a stored sweep breaks the line instead of receiving an estimate.
Worked example — BTC, $100,000 per leg
On $100,000 of matched notional on each leg, the widest executable BTC pair pays $375 over 7 days before costs. The pair is a long position on dYdX and a short position on BloFin.
The arithmetic multiplies the 1-hour spread of 0.00223% by 168 hours and by $100,000. It holds the rate at the sweep value of 09-09 11:52 UTC. Real funding resets at every settlement interval, so this figure is an illustration of size, not a forecast.
Net opportunity after costs: incomplete
A tradable estimate needs entry and exit fees, spread and slippage on BOTH legs (from our order-book cost engine), holding-period assumptions and funding-interval uncertainty. Until every one of those inputs is wired into this page and documented in the methodology, we show the honest label instead of a precise-looking net number: incomplete. Cost-adjusted analysis ships after the history series matures (preliminary · 288 sweeps stored).
Four inputs stay missing today. The taker fee applies on both legs at entry and again at exit. The quoted spread and the slippage apply on both legs at the traded size. The funding drift between sweeps changes the collected rate before the first settlement. The borrow cost of the collateral applies for the full holding period. PerpFinder publishes a net number when it can measure each of those four.
Risks in a funding pair
Funding arbitrage is not risk-free. Each venue resets its funding rate at every interval, so a wide spread can close before the first settlement. A sharp price move can liquidate one leg while the other leg stays open, which turns a market-neutral pair into a directional position. Venue risk applies to each account separately, because withdrawal limits, regional access rules and platform failure reach one side without reaching the other.
Collateral also costs money. Margin on one venue cannot support a position on the second venue, so the pair ties up capital twice. A venue that changes its funding interval changes the normalized rate that this page compares, and the methodology states which intervals PerpFinder reads and which it assumes.
The funding rates page holds the full venue table with current, predicted and realized rates. The history page tracks the maturity of the stored series. The cost comparison tool measures order-book execution cost, which this page needs before it can publish a net figure. The stored sweeps of the last 30 days are available through the public funding history API.
Questions about funding divergence
What is cross-venue funding divergence?
Cross-venue funding divergence is the gap between the funding rates of two perpetual venues for the same asset in the same sweep. PerpFinder converts every venue rate to a one-hour basis before it subtracts, so venues with 1-hour, 4-hour and 8-hour intervals stay comparable. The original venue rate and interval stay unchanged in storage.
Why is an annualized funding spread not an expected return?
The annualized column extrapolates one observed hour across 8,760 hours. Funding changes at every settlement interval, and a wide spread often closes within hours. The number ranks venue pairs against each other on one scale. It does not forecast a yearly result, and PerpFinder never presents it as one.
What does executable mean on this page?
A venue pair is executable when both legs reported open interest of $1,000,000 or more in the same sweep. The floor removes venues that cannot carry a position of useful size. It is a size filter only. It is not a measurement of order-book depth, and it is not a promise of fill quality.
How often does the funding sweep run?
The sweep reads every covered venue every 15 minutes and stores the result. Collection started on 2026-07-16. The tables show the most recent stored sweep with its UTC timestamp, and the chart summarizes the last 30 stored days.
Why does this page show no net result after costs?
A net figure needs the trading fee on both legs at entry and exit, the spread and slippage on both legs, the funding drift over the holding period, and the borrow cost of the collateral. PerpFinder does not have every one of those inputs wired into this page. Until each input is wired and documented in the methodology, the page shows the gross spread and the label incomplete.
PerpFinder Research
Editorial TeamEditorial team tracking 100+ perpetual futures venues with live on-chain and exchange data.