Why Do the Same Asset's Funding Rates Differ Across Exchanges?.

The same asset prints different funding rates on different exchanges because funding is not a property of the asset. It is a property of each venue’s market in that asset: its own positioning, measured by its own formula, against its own index, on its own clock. Bitcoin does not have a funding rate. Bitcoin perps on Hyperliquid, Binance, and Variational each have one, and the four structural differences between venues guarantee the numbers diverge even when nothing interesting is happening.

First, positioning differs by venue. Funding measures which side of a specific market is crowded, and each venue hosts a different crowd: one exchange’s user base can be net-long the same asset another exchange’s users are net-short, and each venue’s rate faithfully reports its own imbalance. This is the deepest reason, because it means cross-venue funding differences are real information about where different kinds of traders sit, not measurement error.

Second, the measurement machinery differs. The formula skeleton is shared, per how funding is calculated, but every parameter around it is a venue choice: index composition, impact notional, sampling cadence, five seconds on Hyperliquid against sixty on Variational, flat averaging against the recency weighting covered in the recency-weighted funding page, and the caps from the caps and clamps page. Two venues running the same market with identical positioning would still print slightly different rates on parameters alone.

Third, the clocks differ, and this one distorts comparisons more than any other. A venue settling hourly and a venue settling every eight hours are not quoting comparable numbers even when the digits match, an eightfold difference in carry hiding behind identical displays, the normalization problem from how often funding is paid. Any cross-venue comparison that has not converted everything to a common per-hour or annualized basis is comparing noise.

Fourth, the structure of liquidity differs. An order-book venue’s premium emerges from resting orders and taker flow; a dealer venue like Variational’s Omni, where a single liquidity provider quotes every market and hedges externally, produces its premium through the dealer’s quoted prices, a mechanism I covered in the funding rate with no order book. Different liquidity structures respond differently to the same flow, especially in fast markets and on the long tail.

Which raises the obvious question: if rates differ, why doesn’t arbitrage flatten them? It does, partially and constantly, and the residual gap is the honest measure of the costs and risks of the trade that would close it: capital split across two venues that cannot share margin, fees and slippage on both legs, settlement clocks that pay the two legs at different times, and the divergence risk covered in why delta-neutral positions still get liquidated. The spread that survives all of that is exactly the spread you see, which is why a persistent cross-venue funding gap should be read first as a price on the friction between venues and only second as free money. The full honest math on harvesting it is in what funding rate arbitrage is and whether it works.