FACTA

tape.pool-check

Version
0.1.0
Terms named on this page

How much of a pool's reported volume over one window came from traders who bought and sold back the same amount, how its volume compares with the liquidity it traded against, and the Pool Check term that follows from those figures by fixed thresholds.

Changelog

Changelog
VersionDateChange
0.1.0
2026-11-17
First version, with the spec's default thresholds. They are to be calibrated on the first 100 pool-days; a change of any threshold is a new version.

What it measures

Reported volume on this chain can be many times the liquidity it traded against. This method takes every swap of one pool over one window and measures who traded, how much of the volume returned to where it started, how concentrated it was and how it compares with the pool's in-range liquidity. It reports what the swaps show, against which threshold.

It does not measure why anyone traded, who a trader is, or whether a price was fair. A high round-trip share says that traders bought and sold back; it says nothing about their intent. It does not measure the venues Stock Tokens also trade on outside Uniswap: RFQ, a proprietary AMM and an order book. A pool page says so.

Inputs

For one pool and one window, here one UTC day:

  1. Every Swap event of the pool in the window. Uniswap v3: Swap(address,address,int256,int256,uint160,uint128,int24) emitted by the pool, topic 0xc42079f94a6350d7e6235f29174924f928cc2ac818eb64fed8004e115fbcca67. Uniswap v4: Swap(bytes32,address,int128,int128,uint160,uint128,int24,uint24) emitted by the PoolManager (0x8366a39cc670b4001a1121b8f6a443a643e40951) with the pool ID as its first topic, topic 0x40e9cecb9f5f1f1c5b9c97dec2917b7ee92e57ba5563708daca94dd84ad7112f. Read with eth_getLogs, 2,000 blocks a request, every pool of a run in one request per venue.
  2. The transaction of each swap (eth_getTransactionByHash): its sender is the trader. A swap event names the contract that called the pool, which is usually a router many people use, not the trader.
  3. The receipt of each transaction sent to an ERC-4337 entry point (v0.6 0x5ff137d4b0fdcd49dca30c7cf57e578a026d2789, v0.7 0x0000000071727de22e5e9d8baf0edac6f37da032, v0.8 0x4337084d9e255ff0702461cf8895ce9e3b5ff108). Such a transaction is a bundle sent by a bundler; the trader of a swap inside it is the smart account of the user operation that holds it, the sender of the first UserOperationEvent after the swap.
  4. Block times of the window's first and last block, which give its length.

The direction of each amount, by venue. On Uniswap v3 an amount above zero went into the pool; on Uniswap v4 an amount below zero did, because the event reports the swapper's side. Both were checked on this chain's logs on 2026-11-17: within a pool, every swap after which the price of token0 fell had put token0 into the pool (3,197 of 3,197 on v3, 402 of 402 on v4).

The quote asset is the pool's USDG or WETH side. Every figure in a quote asset is in that asset's own units: a WETH pool's volume is in WETH, never converted to dollars. A pool with no USDG or WETH side, or a StablePair pool between two issuer-listed tokens, is not computed.

Liquidity is read from the events. No public endpoint of this chain serves state older than its head block, so the pool's liquidity at the start and the end of the window is the in-range liquidity and price that the window's first and last Swap events report, right after each of them.

The figures

The figures
FigureDefinition
swaps
Swap events of the pool in the window
traders
Distinct traders
reported_volume
Sum of every swap's notional: the quote asset paid in or taken out
round_trip_volume
For each trader whose buys and sells net to under 5% of its gross, twice the smaller of the two
clean_volume
Reported volume less round-trip volume
round_trip_share
Round-trip volume ÷ reported volume
shared_contract_share
Share of reported volume whose swaps were called by a contract that two or more traders used in the window. Many people can sit behind one such contract, which is why the trader is taken from the transaction
concentration_top5
Share of reported volume from the five traders with the most volume
bot_share
Share of reported volume from traders with more than 200 swaps in the window
liquidity_value_start, liquidity_value_end
In-range liquidity value right after the window's first and last swap
turnover_per_day
Reported volume ÷ the mean of the two liquidity values ÷ the window's length in days
largest_swap, median_swap
The largest swap's notional, and the median swap's
largest_to_median
Largest swap ÷ median swap
largest_to_liquidity
Largest swap ÷ the mean of the two liquidity values
window_seconds
From the first block's time to the last block's, both counted

In-range liquidity value, and how it differs from total value locked. A pool reports its in-range liquidity L and its price √P. The value, in the quote asset, of a full-range position as deep as the pool is at that price is 2·L·√P when the quote asset is token1 and 2·L/√P when it is token0. This method uses that value. It is not the total value locked: tokens deposited outside the current price range are not in it, and a position concentrated around the price counts at the depth it gives, which can be more than the tokens it holds. Sites that report "liquidity" usually report the total value locked, so their turnover differs from this one.

How the median is kept. Raw swaps are never stored. Each swap's notional is counted in a bucket, 32 buckets for each doubling, found with whole numbers only; the median is the middle of the bucket holding the middle swap. It is within about 1.1% of the exact median.

Every share and ratio is computed with whole numbers, in millionths (six decimals) and turnover in thousandths a day (three decimals), rounded down. Every threshold compares the same whole numbers, so every machine that reads the same swaps gets the same term.

The term

The Pool Check is one term for each pool and window, by the first test that applies:

The term
TermTest
Insufficient data
Fewer than 20 swaps in the window
Recycled
Round-trip share above 50% (500,000 millionths), or turnover above 50 a day, or one swap larger than the mean in-range liquidity value
Organic
Round-trip share under 15% and turnover under 10 a day
Mixed
Everything else

A window carries the tests that decided its term, in this order: "Fewer than 20 swaps in the window", "Round-trip volume is above 50% of reported volume", "Turnover is above 50 a day", "One swap is larger than the pool's in-range liquidity value", "Round-trip volume is under 15% of reported volume and turnover is under 10 a day", "Neither the Recycled test nor the Organic test is met". A Recycled window lists every Recycled test it meets.

The evidence each figure carries

Every figure names the window's blocks and cites the filter that reads all of the pool's swaps over them: the emitting contract, the Swap topic and, on v4, the pool ID, with the number of logs it returned. The two liquidity values also cite the swap they were read from, and largest_swap the largest swap. Anyone can send the same eth_getLogs and get the same logs.

Recomputation

pnpm tool packages/tape/recompute.ts --window <window.json> --log-rpc <endpoint> --rpc <endpoint> reads the window's swaps again, resolves their traders, computes every figure with the same code and prints match or mismatch for each one, and for the term. It needs an endpoint that serves eth_getLogs over old blocks.

Known ways to game the method

  • Many wallets. A round trip split across wallets, one buying and another selling, is not one trader's round trip: this method sees two traders with a net position each. Concentration and the share of traders with more than 200 swaps are reported beside it for that reason, and a later version may link wallets that fund each other.
  • Liquidity added for the window. Liquidity added just before the first swap and removed after the last makes the pool look deeper at both ends than it was in between. The two liquidity values are shown, with the swaps they were read from.
  • Waiting for the window to close. Trades kept just under a threshold, or split across two days, are measured as they fall. Every threshold is public, and so is every figure beside the term.
  • A contract that serves one trader. Calling the pool through one's own contract leaves shared_contract_share low; calling it through a public router raises it. The figure describes where the swaps came from and nothing else.

Limits

  • Uniswap v3 and v4 pools only, with a USDG or WETH side. Stock Tokens also trade through RFQ, a proprietary AMM and an order book, which this version does not read.
  • Hooks. On Uniswap v4 a hook can take or add amounts of its own around a swap. The Swap event reports the pool's own swap, and so does this method.
  • A relayer or a solver that sends transactions for many people is one trader here.
  • The window is one UTC day, computed after the day has ended, in the batch lane. A pool's page shows its newest window and the days before it; nothing is computed while a caller waits.
  • A pool with no swap in a day has no window for that day.

What else exists

Analytics sites report volume and liquidity for these pools from the same events, usually against the total value locked. Some flag wash trading with their own models, which are not public. This method publishes its thresholds, its figures and a script that recomputes them, and it reports a term from a fixed list rather than a score.