<100 subscribers
The 2-Second Window That Paid 7 Figures
HyperEVM blocks tick every 2 s; Hyperliquid spot ticks every 250 ms.
That 1.75 s delta is the world’s simplest edge: HYPE can print “stale” on-chain while the off-chain order-book has already moved.
We turned the gap into a 24/7 assembly line: buy the lag, sell the real price, hedge flat, repeat.
Week 0 – MVP in 48 h
One Python loop, two RPC endpoints.
Trigger: 15 bp spread after fees.
First day: 200 k USD volume, +400 USD PnL.
Two competitors showed up; we decided to starve them.
The Fee-Rebate Kill-Shot
April: Hyperliquid introduced HYPE-staking rebates.
We mortgaged the flat, bought 100 k HYPE, locked it, captured 30 % maker-discount.
Breakeven spread dropped from 15 bp → 5 bp.
Competitors’ maths broke; we pushed volume to 500 M in two weeks and watched them power-down their bots mid-flight.
Scaling Walls & How We Bulldozed Them
Gas ceiling – 2 M/block, ~8 arb-txs max.
→ 120 fresh wallets, round-robin submission, dynamic ROI filter.
Taker fatigue – 2.45 bp lost per leg.
→ Became maker on Hyperliquid, quoting 1-tick wide, saving fee + capturing wicks.
Stable-coin slippage – USDT₀ (EVM) vs USDC (Hyperliquid).
→ Skip the swap: keep dual-treasury, rebalance only when delta > 1.2 M USDC or 300 k USDT₀.
Perp untouched – funding 5–15 bp/8 h, basis 3–8 bp.
→ Added delta-neutral perp leg, collected 600 k USD in funding alone.
Risk Stack – No Luna-Style Surprises
Max block-long inventory: 80 k HYPE (≈ 5 M USD).
Auto-liquidate if delta > 100 k HYPE for 45 s.
Progressive ROI: bigger clip → higher hurdle.
Zero socialised leverage; every position is fully hedged in the next block.
Team of Two – The Only Org-Chart We Needed
Brother A: codes, dev-ops, infra.
Brother B: parameter tuner, liquidity scout, “human circuit-breaker”.
Daily ritual: 08:00 UTC joint Zoom, walk through fills, skew, open interest; push at most three knob-twists per day. No Slack, no Jira, no governance token—just a shared Notion board and a 2-of-3 Gnosis safe.
The Numbers That Shut Our Robot Down
8 months
125 B USD two-way volume (≈ 5 % of Unit chain lifetime)
5 M USD net, after 1.2 M gas and 300 k funding payments
2 000 + hrs dev time → effective hourly wage: 2 500 USD
We have officially pulled the plug. Bot is sleeping; capital is back in stables. Season 3 incentives are rumoured to re-price the entire curve. We’ll be in the air from Istanbul to Dubai again—this time just for the coffee.
TL;DR for Copy-Cats
Speed is commoditised; cost structure is destiny.
Rebate engineering > alpha discovery.
Every 1 bp you save on leg #1 is 1 bp you can give away to starve competitors.
If you can’t draw the risk diagram on a napkin, don’t deploy.
When the arb delta closes, stop trading—start building the next edge.
The 2-Second Window That Paid 7 Figures
HyperEVM blocks tick every 2 s; Hyperliquid spot ticks every 250 ms.
That 1.75 s delta is the world’s simplest edge: HYPE can print “stale” on-chain while the off-chain order-book has already moved.
We turned the gap into a 24/7 assembly line: buy the lag, sell the real price, hedge flat, repeat.
Week 0 – MVP in 48 h
One Python loop, two RPC endpoints.
Trigger: 15 bp spread after fees.
First day: 200 k USD volume, +400 USD PnL.
Two competitors showed up; we decided to starve them.
The Fee-Rebate Kill-Shot
April: Hyperliquid introduced HYPE-staking rebates.
We mortgaged the flat, bought 100 k HYPE, locked it, captured 30 % maker-discount.
Breakeven spread dropped from 15 bp → 5 bp.
Competitors’ maths broke; we pushed volume to 500 M in two weeks and watched them power-down their bots mid-flight.
Scaling Walls & How We Bulldozed Them
Gas ceiling – 2 M/block, ~8 arb-txs max.
→ 120 fresh wallets, round-robin submission, dynamic ROI filter.
Taker fatigue – 2.45 bp lost per leg.
→ Became maker on Hyperliquid, quoting 1-tick wide, saving fee + capturing wicks.
Stable-coin slippage – USDT₀ (EVM) vs USDC (Hyperliquid).
→ Skip the swap: keep dual-treasury, rebalance only when delta > 1.2 M USDC or 300 k USDT₀.
Perp untouched – funding 5–15 bp/8 h, basis 3–8 bp.
→ Added delta-neutral perp leg, collected 600 k USD in funding alone.
Risk Stack – No Luna-Style Surprises
Max block-long inventory: 80 k HYPE (≈ 5 M USD).
Auto-liquidate if delta > 100 k HYPE for 45 s.
Progressive ROI: bigger clip → higher hurdle.
Zero socialised leverage; every position is fully hedged in the next block.
Team of Two – The Only Org-Chart We Needed
Brother A: codes, dev-ops, infra.
Brother B: parameter tuner, liquidity scout, “human circuit-breaker”.
Daily ritual: 08:00 UTC joint Zoom, walk through fills, skew, open interest; push at most three knob-twists per day. No Slack, no Jira, no governance token—just a shared Notion board and a 2-of-3 Gnosis safe.
The Numbers That Shut Our Robot Down
8 months
125 B USD two-way volume (≈ 5 % of Unit chain lifetime)
5 M USD net, after 1.2 M gas and 300 k funding payments
2 000 + hrs dev time → effective hourly wage: 2 500 USD
We have officially pulled the plug. Bot is sleeping; capital is back in stables. Season 3 incentives are rumoured to re-price the entire curve. We’ll be in the air from Istanbul to Dubai again—this time just for the coffee.
TL;DR for Copy-Cats
Speed is commoditised; cost structure is destiny.
Rebate engineering > alpha discovery.
Every 1 bp you save on leg #1 is 1 bp you can give away to starve competitors.
If you can’t draw the risk diagram on a napkin, don’t deploy.
When the arb delta closes, stop trading—start building the next edge.
Share Dialog
Share Dialog
No comments yet