What Is Funding Rate Arbitrage, and Does It Actually Work?.

Funding rate arbitrage is holding offsetting perp positions on the same asset across two venues, long on the venue where the funding flow favors longs and short on the venue where it favors shorts, so the price exposure nets to zero while both legs collect or minimize funding. If asset X has deeply negative funding on venue A, where longs get paid, and positive funding on venue B, where shorts get paid, a long on A against an equal short on B earns both flows. Does it work? Yes, conditionally: it is a real strategy that professional desks run at scale, and the conditions are exactly the things retail screeners omit.

The steelman deserves a fair statement, because the trade’s logic is sound. The position is market-neutral by construction, the yield source is a genuine structural flow rather than a directional view, and persistent cross-venue spreads exist for the reasons in why funding differs across exchanges: segmented crowds, different parameters, costly friction between venues. The arbitrageur is being paid to supply something real, the capital and operational tolerance to bridge two markets that cannot bridge themselves.

Now the honest math, which is mostly subtraction. From the gross spread, subtract trading costs on four executions, entry and exit on both legs, plus slippage that scales with size on the thinner leg. Subtract the capital drag of posting margin on two venues that cannot see each other, which halves your effective leverage. Subtract the interval mismatch when the legs settle on different clocks: the spread you measured at entry is not the spread that pays, since each leg’s rate keeps moving until its own snapshot, per how often funding is paid. And subtract an allowance for the tail: the divergence-and-liquidation scenario from why delta-neutral positions still get liquidated, which converts months of basis-point harvesting into one bad hour. What survives the subtraction on a typical retail-sized, screener-flagged opportunity is frequently nothing, the arithmetic of the 1,500% APR that doesn’t exist.

The conditions under which it genuinely works are knowable in advance. Same or aligned settlement intervals on both legs, so the spread cannot drift between mismatched snapshots. Real depth on both legs at your size, which on the long tail is the binding constraint per the low-volume problem. A spread that has demonstrated persistence rather than appeared this hour, since a spread that has survived days of reversion pressure is information while a fresh one is overwhelmingly likely to be gone before your first collection, the argument from funding rate mean reversion. Modest leverage with a liquidation buffer sized to survive the divergence you have not imagined yet. And costs measured honestly at your actual size, not at the screener’s frictionless fantasy size.

Run through that checklist and the trade transforms from the screener’s version, fifty opportunities a day at four-figure APRs, into the real version: a few setups a week at modest annualized yields that compound respectably and occasionally gap against you. That real version is worth doing for the operationally serious. The fantasy version funds the people selling it, which is precisely why the screener I built surfaces interval alignment and persistence instead of headline APRs. How this trade differs from its sturdier cousin is in cash and carry versus funding arbitrage.