Funding rate divergence between exchanges on the same perp i’ve been obsessing over this for months

Basically any time i pull up the same perp contract on two different exchanges the funding rates are different in ways that don't obviously explain themselves, and it's been bugging me enough that i started tracking it systematically.

last week's example: ETHUSDT funding on bitmex was averaging +0.012% per 8h for three days while on bybit over the same window it was +0.034%. same underlying, similar OI profiles, nearly identical mark prices.

that's a pretty wide persistent gap, enough that a delta-neutral carry trade between the two would have printed real money if you could execute it without too much friction.

The obvious explanations i keep hearing are things like regional taker flow pushing inventory imbalances different directions on different exchanges or differences in how each one actually calculates the funding formula (the clamp ranges vary, premium component weighting varies), or taker flow skew that doesn't show up in public OI data. i buy all of those as partial explanations. none of them really account for why the gap persists for days instead of getting arbed to equilibrium within hours.

if this spread is visible to me from retail-tier tooling, serious trading desks have to be seeing it too.

so either they're not running the trade and i'm missing a reason why, or they're running it and the persistent gap IS the equilibrium price of whatever the friction is. capital costs on both sides, transfer latency, jurisdictional constraints, the bitmex trust factor for parking serious size on the offshore side vs newer exchanges – all plausible but i can't verify.

If anyone's actually running cross-venue funding carry at real size i'd love to know what the edge looks like after you account for funding the collateral on both legs or if it turns out what looks like alpha from the outside is just a mirage once you try to execute it