Funding Arb 101 — The Oldest Delta-Neutral Play in the Wasteland
// The tourists trade price. You trade the difference in what two venues charge each other for the same position. Here's the mechanic, the math, and the trap that kills the tourists.
Funding rate arbitrage is the oldest delta-neutral play in the wasteland. The tourists trade price, which means they bleed every time the chart turns; you trade the difference in what two venues charge each other to hold the same position, and the chart becomes noise. Here's the mechanic, the math, and the trap that kills the tourists.
One spread, two legs, zero directional exposure. It sounds like magic. It's not. It's a structural inefficiency that shows up every time a new venue opens its books and forgets to price-match the older ones. The edge lives in the gap. The Forge exists to show you where the gap is.
// WHAT FUNDING ACTUALLY IS
A perpetual futures contract has no expiry. Nothing forces its price back to spot. So every venue invents a funding mechanism — a payment, exchanged between longs and shorts, that drags the perp mark back toward the underlying. When the longs dominate, longs pay shorts. When shorts dominate, shorts pay longs. The payment is quoted as a rate and applied per settlement interval.
Read that again. It's not the exchange taking fees. It's the crowd paying itself, mediated by a mechanism designed to punish whichever side got too loud.
That mechanism is the edge.
// SEE TODAY'S FUNDING RATES// WHERE THE ARB LIVES
Different venues run different books. Different books carry different longs-versus-shorts imbalances. Different imbalances produce different funding rates — for the same token, at the same moment, across the wire.
BTC on Venue A is paying +0.012%/hr. Longs on A bleed. BTC on Venue B is paying −0.009%/hr. Shorts on B bleed. The spread is 0.021%/hr.
That spread is the lane.
You short BTC on Venue A and start collecting A's +0.012%/hr funding. You long BTC on Venue B and start collecting B's +0.009%/hr funding. Both legs are opposite, so BTC's chart can do whatever it wants — up, down, sideways, ninety-percent drawdown — and your pair stays flat. Every settlement interval, the funding drips yield into both sides of your ledger.
The math doesn't care about BTC's chart. It cares about the spread holding open.
// THE MATH IN BLOOD
Hourly rates annualize cleanly. 0.021%/hr × 24 hours × 365 days = 183.96% APR gross.
That's the naked number. It's also the number the tourists quote when they post screenshots. It's not what you actually earn.
Three things carve into it:
Taker fees — four times. Every leg is entry plus exit. Two venues, two entries, two exits — four taker bills. Bleed this on both ends.
Slippage — twice. Each leg crosses the spread once on entry. Thin books eat more than thick ones.
The hold window. Costs are fixed per round trip. Amortised over a 7-day hold, a 20bps round-trip is worth about 104% APR in break-even. Hold it 30 days, same cost is 24% APR. Hold it 24 hours, it's 730% APR. Short holds die fast.
Net APR is what you keep. Gross is vanity. The Forge shows both — net is king.
// OPEN THE COST LEDGER// THE EXECUTE
Four steps. Do them in order. Don't skip the third one.
One — pick the pair. On the board, sort by NET APR descending. Pick a pair with positive net APR after fees amortised over your intended hold. Click the card — the strategy modal opens with gross APR, fees, the gross-to-net waterfall, and 24h of history. If the waterfall's net bar is thin or negative, walk away.
Two — size the legs. Same notional on both sides. Not the same token count — the same dollar value. If BTC is $75k and you're running $10k, that's 0.133 BTC long on Venue B, 0.133 BTC short on Venue A. Delta-neutral means equal-notional, not equal-unit.
Three — enter both legs in the same minute. This is where the tourists die. They fire Venue A, then wait for "a better entry" on Venue B, then watch BTC rip 3% against them and eat a $300 leg-risk bill before the hedge is on. Fast. Same. Minute.
Four — hold, collect, exit together. Every funding settlement adds to your ledger. When the spread compresses — and it always compresses, because the tourists eventually notice — you close both legs in the same minute. Same discipline going out as coming in.
// BACKTEST THE PLAY// WHAT KILLS THIS PLAY
Three things eat you. Know them before you ape.
Funding flips. The spread that opened can close, and then invert. If Venue A's rate drops below Venue B's mid-hold, the payment flow reverses and suddenly you're paying. Set the alert wire. Watch for the flip. Close early if the edge dies — a flat trade is not a reason to keep holding.
Liquidation on one leg. The pair is delta-neutral, but each leg has its own margin account. If BTC rips hard enough on either venue to liquidate one side — because a venue froze, a margin add failed, or you aped the max multiplier — the hedge dies. What you have left is directional exposure you didn't sign up for, at the worst possible moment: price has already moved, your stop is nowhere near, and the other leg is now a naked bet. Keep both legs over-collateralised. Don't run max multiplier on a funding carry. The edge is 50-200% APR at best. Size like you're keeping the money, not like you're gambling for a screenshot. Margin blows are permanent. Spreads are forever.
Venue death. Drift got hit. Bluefin's public API returned 503 for weeks. Vertex flakes on the hour. When a venue dies, the leg on that venue dies with it — margin locked, no exit, the hedge half-up in smoke. Diversify venues across multiple plays. Don't concentrate both legs of the same play on two thin chains.
// WHEN TO TAKE THE TRADE
Not every spread is worth running. Three gates before you size up.
Net APR ≥ 20% at your target hold. Anything lower and you're racing cost amortisation. The tourist traps sit at 5-10% net — they look juicy until the fees bleed them flat.
Both venues have real open interest on that symbol. A wide spread on a pair where one venue has $50k of open interest means the book will flip the moment you size in. Read the OI column. Thin OI on one side = the spread isn't real, it's a pricing error.
The spread has held for at least 24 hours on the historical chart. Click the card, open the modal, check the sparkline. A spread that just appeared from nothing fifteen minutes ago is also fifteen minutes from closing. A spread that's held phosphor-steady for a day is load-bearing.
// READ THE HEATMAP
// WHERE TO HUNT ON THE FORGE
The board is the scanner. The heatmap is the spread atlas. The backtester is the paper trial. The alert wire is the graveyard shift watch.
Every play starts with one of those four tools. Pin the tokens you hunt so they anchor the watchlist. Set the alert threshold at a number that actually matters — 20% net APR minimum, nothing lower is worth a phone buzz. Backtest the pair across the longest window the Forge has stored before you size up. Open the cost waterfall inside the strategy modal and read what the fees actually do to the spread — ninety percent of "juicy" opportunities flatten into noise once you carve the bps out.
The Forge of the Day banner at the top of the board shows the single highest net-APR play on the wire right now, refreshed hourly. Start there. If the featured play doesn't pass your gates, drop to the next card.
The arsenal is free. The discipline is yours to bring.
// SCAN THE BOARD// THE CLOSING DISPATCH
Funding arb is boring. That's why it still pays. The tourists chase 100x screenshots on twitter. The warlords sit on delta-neutral spreads and reap.
One pair. One hold. One ledger. Repeat.
// ENTER THE FORGE →