DEX Liquidity Sniper Playbook

Why liquidity sniping still works

Mempools are still wonderfully noisy. Retail bots spam routers with partially signed bundles, and the majority of them leak intent milliseconds before a block builder commits. If you can sort the noise from the signal, you get three gifts:

1. Early conviction on where the next 30-second trend will start.
2. A live read on how much size is actually willing to cross the spread.
3. A confidence score on whether a venue is about to run out of resting size.

Minimum viable toolkit

• Streamed mempool feed from at least two builders (we watch both MEV-Boost and Blocknative).
• Quote normalizer that converts all pools into USD notional with depth buckets (1k, 5k, 25k...).
• Execution bridge that can fire limit+fill-or-kill orders on-chain from a single command to avoid UI drag.

The playbook

1. Tag every intent with estimated slippage if it executed immediately on the target pool.
2. Fire “ghost quotes” (resting 0.01 token orders) to confirm the venue is actually alive. If two ghosts never confirm, skip the pool.
3. Pre-fund the bridge so you never wait for L1 confirmations. We keep 2% of the desk float on every target chain.
4. Enter in three clips: the first to occupy the top of book, the second to fill right behind the whale, the third to fade if the whale cancels.
5. Set the exit before the fill. Our rule: if slippage doubles from the initial read, auto-exit to the next venue.

Risk notes

• Do not chase a mempool surge if the reactive gas price exceeds 1.8x your base setting.
• Always subtract MEV burn from your expected spread; builders will rebate their friends, not you.
• Document every miss. Nine times out of ten it is because we sized too big for the venue depth, not because the read was wrong.

Thanks to this discipline we average 23 bps edge on TON, 15 bps on Base, and still occasionally catch full percentage point pops when we see a multi-million OTC clip leak into Uniswap.