Hyperliquid Funding Rate Arbitrage vs Binance & Bybit

Hyperliquid pays funding every hour; Binance and Bybit settle every eight. That single mismatch is the whole trade. When Hyperliquid's hourly rate annualizes to 40% while Binance sits at 11% on the same asset, you short the expensive book, long the cheap one, and collect the spread with no directional exposure. The mechanics are simple. The execution — matching legs across a chain and two order books, sizing the hedge so neither side liquidates, and timing entries around three different settlement clocks — is where the edge actually lives or dies.

Why the spread exists in the first place

Funding is a rebalancing payment, not a fee the exchange keeps. On a perp, longs pay shorts (or vice versa) to drag the mark price back toward the index. Each venue computes it independently from its own order-book imbalance and its own premium formula, so the rates diverge constantly.

Three structural reasons the spread persists on Hyperliquid specifically:

• Hourly settlement vs 8-hour on CEXes. Hyperliquid pays 1/24th of the daily rate every hour. Binance and Bybit pay three chunks a day. When a directional move hits, Hyperliquid's rate reacts and settles faster, opening a transient gap that lasts hours, not minutes.
• Thinner books on the long tail. For majors like BTC and ETH the cross-venue spread is usually 5-15% annualized. On mid-caps — think a freshly listed perp — Hyperliquid funding can run 60-120% annualized while the CEX equivalent barely moves, because the on-chain book is dominated by degen longs.
• Capital friction. Bridging USDC to Hyperliquid takes minutes and costs gas. That friction keeps the spread from arbing to zero, which is exactly why it's collectable.

Your gross yield is the funding differential. Your net is that minus fees, minus bridge/borrow costs, minus whatever slippage you eat matching the two legs. If you want the raw API mechanics for reading these rates programmatically, the Hyperliquid API trading bot guide covers the /info endpoints and websocket subscriptions you'll build on top of.

The delta-neutral structure

Pick a direction based on which venue is more expensive to be long on.

Say ETH funding is +0.05%/hr on Hyperliquid (annualizes to ~44%) and +0.011%/hr equivalent on Bybit (~9.6% annualized). Hyperliquid longs are overpaying. So:

• Short ETH-PERP on Hyperliquid → you receive the rich 44% funding.
• Long ETH-PERP on Bybit → you pay the cheap 9.6% funding.

Net carry: roughly +34% annualized on the matched notional, and your net ETH delta is zero. Price can go anywhere; the two perp legs cancel. You're purely harvesting the funding differential.

The critical constraint: notional must match, not margin. If you short 10 ETH on Hyperliquid you long exactly 10 ETH on Bybit. The residual delta between fills is your entry basis — log it, because it's a real cost you'll pay back on unwind.

Hedge sizing so neither leg liquidates

This is the part naive implementations get wrong. Both legs are leveraged perps, and a price move that helps one hurts the other's margin. If ETH rips 20% up, your Hyperliquid short bleeds margin while your Bybit long gains it — but the gain sits on a different venue and can't defend the short in time.

The rule I use: size each leg so a 40% adverse move doesn't liquidate the losing side. That means effective leverage around 2.5x per leg, even though both venues let you run 10-20x.

Worked example on $200k of capital:

Total capital:        $200,000
Split per leg:        $100,000 margin each
Target notional:      $250,000 per leg  (2.5x leverage)
Asset:                ETH @ $2,000  →  125 ETH per leg

Hyperliquid short:    125 ETH, $100k isolated margin
   liq price ≈ $2,000 × (1 + 1/2.5 × 0.9) ≈ $2,720  (+36%)
Bybit long:           125 ETH, $100k isolated margin
   liq price ≈ $2,000 × (1 - 1/2.5 × 0.9) ≈ $1,280  (-36%)

Run a keeper that polls both account states every 15-30s. When either leg's margin ratio drops below 150% of maintenance, auto-transfer USDC to top it up. Budget 5-15 minutes for the Hyperliquid bridge — that latency is the single biggest operational risk, so pre-position a buffer of idle USDC on both venues rather than bridging reactively during a spike. The liquidation-bot clearinghouse writeup explains how Hyperliquid's liquidation engine actually pulls the trigger, which is worth understanding before you set those thresholds.

Isolated margin per leg, always. Cross-margin lets one bad position drain the whole account, which defeats the point of a hedged book.

Settlement timing — the part that separates carry from noise

Because Hyperliquid pays hourly and the CEX pays every 8 hours (00:00, 08:00, 16:00 UTC on Binance/Bybit), the two clocks matter for entry and exit.

• Enter after a CEX funding settlement, not before. If you open the Bybit long 20 minutes before 08:00 UTC and funding there is negative for longs, you eat a payment almost immediately for carry you haven't earned. Open just after a settlement so the first CEX payment is a full period away.
• Hyperliquid funding accrues continuously and settles on the hour. You start collecting the rich short-side funding within the first hour regardless, which is why the Hyperliquid leg is the one you want working for you fastest.
• Watch the funding-flip risk. Hyperliquid rates can invert in minutes when sentiment turns. Track the 8-hour rolling average of the spread, not the instantaneous rate. Only stay in while the smoothed differential clears your fee hurdle. If it compresses below breakeven for more than a couple hours, unwind.

Fee math and breakeven

Hyperliquid: 0.02% maker / 0.05% taker. Bybit and Binance perps: ~0.02% maker / 0.055% taker. Post-only both entries and you're paying roughly 0.02% × 4 legs = ~0.08% round-trip across the pair on notional.

On $250k notional that's ~$200 in fees. At a 34% annualized net spread, you earn ~$232/day on $250k. So you clear fees in under a day and everything after is profit — as long as the spread holds. The trades that lose money are the ones where the spread compresses to 8% right after you pay 0.08% to get in; always require the smoothed spread to be at least 3-4x your round-trip fee before opening.

What actually breaks

• Spread mean-reverts on entry. You paid taker fees chasing a fill, the rate normalized an hour later, and you're carrying a 6% net spread you can't cover. Discipline on the entry threshold fixes this.
• Bridge latency during a gap. Covered above — pre-fund, don't react.
• One venue delists or halts the perp. Now you're running a naked leg. Keep position caps per asset and don't concentrate the whole book in one thin mid-cap.
• Funding sign flips on the leg you're receiving. Rolling-average monitoring is non-negotiable.

If you're already running directional strategies, the funding-arb book pairs well with a Hyperliquid perps execution bot for the leg management, and the whole cross-venue P&L is far easier to reason about when funding, basis, and margin ratios sit on one live trading dashboard instead of three exchange tabs. For a related on-chain yield structure that shares most of this plumbing, the HLP vault strategy breakdown is worth a read.

The whole thing is mechanical, not predictive — you're not calling direction, you're renting out the rate differential and managing margin around it. That's what makes it durable, and also what makes the instrumentation, not the idea, the actual work.

If you'd rather have the cross-venue matching, hedge sizing, and margin keepers built and running in production, that's exactly what our Hyperliquid funding arbitrage service is scoped to deliver.