Uniswap Hook Incubator · UHI10 Hookathon

Impact
Rebated Fees

A Uniswap v4 hook that charges every swap a refundable deposit, then uses the pool’s own price to decide who keeps it.

Uniswap Hook Incubator · UHI10

A security deposit
on every swap.

A Uniswap v4 hook that charges every trade a refundable deposit, then uses the pool’s own price sixty seconds later to decide whether to give it back.

Benign · refunded Informed · paid to LPs

The problem

One fee. Two completely different traders.

Someone swapping five hundred dollars pays the same fee as a bot that just took money off the pool. A pool has one fee and no way to tell them apart, so it charges everyone enough to survive the bot.

The small trader pays for damage they did not cause.

The insight

You cannot tell at the time. A minute later it is obvious.

At the moment of a trade there is genuinely nothing to see, because whatever the trader knew has not reached the price yet.

If the price carries on moving the same way afterwards, the pool was still quoting a stale number when that trader left, and it is about to be picked off again on the same mistake.

How it works

01

Everyone pays

The pool charges its normal fee, untouched. The hook takes 25 bps more and holds it as a claim inside Uniswap’s own vault. No judgement, because the next sixty seconds have not happened.

02

The pool keeps count

It records where the swap left the price, then runs a time-weighted average of everything after. A single block cannot move that average.

03

Anyone settles it

Drift past the threshold and the deposit goes to the liquidity providers. Otherwise it returns to the trader. Whoever settles keeps 2%, so self-settling costs nothing.

Evidence · 26,209 real mainnet swaps replayed

refunded to the trader paid to the liquidity providers

Not a simulation. Real trades from the live USDC/ETH pool, replayed to ask what this hook would have decided. 3.2% land past the threshold.

Chosen on economics, not on hit rate

We did not pick the highest revenue setting.

Thirty six parameter combinations were tested against what each one costs a trader carrying no information. The settings with the best headline revenue charge ordinary traders three times as much.

+32.7%
LP revenue over the base fee
0.71 bps
expected cost to an ordinary trader
42.8%
of captured value recovered
$305,032
captured by informed flow that week

How it compares

Everyone else

Decides before the swap

  • Guesses from volatility, or a gap against an oracle
  • Needs a price feed to trust and pay for
  • Overcharges ordinary traders who arrive in a volatile minute

PegGuard · TRIDENT · DAMM · Nezlobin Directional Fee · Arb Controller

This hook

Decides after the swap

  • Reads what the price actually did over the next sixty seconds
  • No oracle, no off-chain service, no keeper network
  • Observes the outcome instead of forecasting it

Zero of 660 prior Hook Incubator projects settle a fee after the fact

The version that failed

The first build flagged fewer trades than random chance.

It measured drift from the price the trader paid. Your own trade moves the price, and that movement bounces back, which drags every trader toward looking innocent.

Measuring from where the swap left the price removes your own footprint, and the signal appears. Found before writing any Solidity. That one choice is the difference between a mechanism that works and one that quietly does nothing.

Honest limits

StatisticalA trade carrying no information still trips the threshold about one time in thirty five. That is exactly where the 0.71 bps comes from.
Clean arbitrageAn arbitrageur who corrects the price in one move is not caught. Nothing keeps drifting, so there is nothing to measure.
ManipulationReads benign, because manipulation is a price move that snaps back.
BacktestThe replay assumes the same trades would still have happened with the deposit in place.

Live on Unichain Sepolia

Every swap posts a deposit.

Sixty seconds later the pool’s own price decides whether you were trading on information, and the deposit either comes back to you or goes to the people whose money you traded against.

26
tests, 96% line coverage
200
cases cross-checked against the analysis
0
partner integrations, by design
0x7dEC15A3…C044
hook address

← → to navigate