Method
Everything here is read from Robinhood Chain's public RPC — the Uniswap v4 PoolManager's own logs. No vendor feed, no API key, no backend. The numbers are a census of the window, not a sample of it.
One window, every pad. Launches are taken from a single block range, and every pool in it has had at least the trailing edge of that range to finish dying before it is counted. This is the whole point of the site: a pad that launched three days ago and a pad that has run for months cannot be compared on all-time numbers, because the comparison measures age.
A coin is alive if somebody is still trading it. Two earlier definitions were tried on a sibling site and both failed: still has liquidity reported 76% of day-old launches as alive, because pads on this chain lock LP at launch and the liquidity outlives the coin; above its launch price reported 98.9%, because a single-sided launch only moves up as people buy. Neither measured anything.
So a lifespan is the time from initialization to last trade, and there is no inactivity rule at all — a pool that traded once at minute one and once at minute ninety lived ninety minutes, gap included. That is deliberate: declaring death after a silent period would make every number on this site a function of how long a silence has to be.
The one time threshold that remains is censoring. A pool whose last trade is very recent might trade again, so its lifespan is a lower bound rather than a completed observation; those are counted and published rather than silently rounded down.
Uniswap v4 has no factory and no per-pool contract, so a pool cannot be traced to its creator by address. Two signals are used, in this order:
CashCatQuoteFeeHook — so it identifies the coin, not the pad.
The order matters and was not guessed. Sampling found that every
pools.trade launch arrives through an ERC-4337 bundler, so the
transaction's to is the EntryPoint rather than the pad. Reading
to first would have filed all of Uniswap's launches under
"EntryPoint" and made the pad vanish from its own leaderboard. Known
infrastructure addresses are excluded by name for the same reason.
Pads set different bars for a launch to reach open trading. Comparing graduation rates across pads would measure the threshold rather than the coins, so no graduation rate appears anywhere on this site. Only outcomes measured identically for every pool are compared.
A sibling site once published a 32-minute half-life that turned out to track its own ten-minute threshold rather than the coins. The equivalent risk here is the survival mark: the headline names the moment two pads are furthest apart, and picking that moment from the data is exactly how you flatter a finding. So the same comparison is run at every sampled mark and the result is published below — if a ranking only holds at the mark that was chosen, it is an artefact and this table says so.
"Traded" means a swap emitted by the v4 PoolManager. A launchpad whose trades settled inside its own hook instead would look, from here, exactly like a launchpad nobody buys from — and the age check below could never catch it, because that error does not change with time. So every hook that emits its own trade event is cross-checked against the PoolManager on every scan, and any disagreement is shown below rather than assumed away.
Fetching a timestamp for every block is impossible at this block rate, so real timestamps are sampled at anchors and interpolated between them. The interpolation is then checked against held-out blocks that were never anchors, and the worst observed error is published with the data rather than assumed away.
The landing page leads with one moment — the mark where two launchpads are furthest apart and their intervals do not overlap. That moment is chosen by searching, which means the honest question is whether the same pair stays separated anywhere else, or only at the point that was hunted for.
| Mark | Headline pair | Separated pairs | of headline pairs |
|---|---|---|---|
| loading… | |||
One failure mode deserves its own test, because the first full scan walked
straight into it. A launchpad whose coins sit in a multi-hour auction before
they reach open trading looks, to this instrument, exactly like a launchpad
whose coins nobody ever wanted: in both cases the pool exists and no swap has
happened. pools.trade's Crowd Launch runs a four-hour bidding
window, and the cohort's trailing edge is closer than that.
So every pad's never-traded rate is computed twice — once for the coins in the older half of the cohort, once for the younger half. If a pad is being caught mid-auction, the younger half must look far worse, because those coins have had less time to finish. If the two halves agree, the auction is not the explanation and the rate is a measurement.
| Launchpad | Older half | Younger half | Difference | Reads as |
|---|---|---|---|---|
| loading… | ||||
The age check above rules out artefacts that change with time. It cannot rule out an artefact that does not — in particular, a launchpad whose trades never reach the PoolManager at all. Uniswap v4 hooks may implement their own trading and emit their own events, so a pad could be busy and invisible here at the same time, permanently, with nothing in the numbers looking odd.
So for every launchpad known to emit its own trade event, each scan compares the pools that event reports against the pools the PoolManager sweep reports. A pool the hook says traded and the PoolManager does not is a pool this site would wrongly be calling dead.
| Hook | Launches | Traded (PoolManager) | Traded (hook's own event) | Invisible | Reads as |
|---|---|---|---|---|---|
| loading… | |||||
Uniswap v3 pools. As of the first scan, v4 carries roughly six times the pool creation of the v3 factory on this chain, and the two populations are not comparable — a v3 launch and a v4 launch are different products with different mechanics. Mixing them would produce one number describing neither.
Prices, market caps and holder counts are also absent. None of them is needed to answer the question this site asks, and each would add a data source that could be wrong in a way nothing here would catch.