Ninety-Eight Per Cent of the Tokens Born on Robinhood Chain Are Dead Within a Fortnight

Robinhood Chain has produced 360,648 tokens since 3 July, about 6,100 every day, across 517,143 trading pools. Of those now between one and two weeks old, 97.7 percent did not trade in the last twenty-four hours. Half of everything the chain makes stops trading before it is twelve hours old. Both figures, and the working behind them, are published from today at the nursery, which is rebuilt every morning.

Launch counts are quoted by every chain and survival rates by almost none, because a launch is one observation and survival takes two. Nobody goes back. So we went back, and this is what the second look found.

Start with what counts as being born, because the choice does most of the work. A token is born here when it first gets a market, not when its contract is deployed. A contract with no pool has no price, nobody who can leave it and nothing that could be said to die, so counting deployments would pack every cohort with things that never lived. Birth is the first time the chain's pool manager names the token, and that moment is exact to the block.

Dying is the harder word and we have deliberately not used it on the page. What is measured is whether a token traded in the last twenty-four hours. A pool sitting untouched is not legally dead, its contract still answers, and in principle somebody could come back tomorrow. In practice almost nobody does, which is the finding.

The decay is far faster than a weekly view can show. Of the tokens now aged between six and twelve hours, 41.3 percent traded yesterday. By one to two days old that is 8 percent. By two to three days, 4.3 percent. By one to two weeks, 2.3 percent. The cliff is not at the end of the first week, where you might look for it. It is at the end of the first day.

Put the other way round, which is the way anyone launching a token would want it: the success rate is 1.55 percent. Of the 309,491 tokens that are more than a week old and have therefore had time to prove themselves, about 4,783 traded yesterday. About about one in sixty-five. Across everything the chain has ever made, including the newest arrivals that have not had their chance yet, it is one in thirty-nine.

The scale underneath those percentages is the part that is easy to lose. The chain is 59 days old and has made more tokens than there are streets in a large city. On 5 August it made 18,795 in a single day and 17,353 the day after, roughly triple its own average, then went back to normal. Whatever happened that Wednesday, it is a spike in the record and it does not appear to have produced anything durable: of the tokens born in that week, 1 percent traded yesterday.

Almost nobody launches a token by hand any more, and the record shows how mechanical it has become. 11,626 different contracts have opened a first pool on this chain. One of them, at `0x4e34…a544`, has put 100,572 tokens into the world by itself, 27.9 percent of everything. A further 192,489 tokens, 53.4 percent of the total, were opened with no hook at all, which is the plain case rather than a platform. The launchpads operating on this chain are known and documented in public; which contract belongs to which of them is not something we can verify from a source we control, so the table gives addresses and says exactly that instead of guessing.

Now the caveat, and it belongs above the fold rather than in a footnote. **The births are a complete count. The survival rates are sampled.** The chain seals a block every 0.101 seconds and its pool manager emits about sixty-six log entries per block, so counting births is possible only because opening a pool is rare enough to filter for. Asking whether each of 360,648 tokens still trades is a different order of work and no public endpoint will answer it. Four hundred tokens are drawn at random from each week and three hundred from each age band, the number checked is printed beside every rate, and the method does not change between runs. A rate from a stated sample is honest. A rate from a sample presented as a census is not.

One band is published and deliberately kept out of the headline. Tokens under six hours old read as less alive than tokens twice their age, which is not a resurrection: the index we ask has often not seen a token born twenty minutes ago, so it reads as dead when it is merely unseen. An earlier draft of this work reported a median life of six hours off exactly that artefact. The figure survives on the page, marked, because hiding a measurement you distrust is worse than showing it with its caution attached.

What is this for. A chain's launch count is used as evidence of life, and on the strength of one it is fair to call Robinhood Chain the most productive token factory in operation. The second number changes what the first one means. 360,648 born and roughly 9,300 still trading is not a busy marketplace, it is a very large machine running at a yield of about one in forty, and anyone deciding where to build, list or buy is entitled to both halves of that before they decide.

It also sets a baseline. This is the first day the figure exists, so there is nothing yet to compare it against, and the nursery will be rebuilt every morning from here on. If the survival rate climbs, the chain is maturing and we will report it. If the daily birth count keeps rising while survival stays at one in forty, that is a different story and we will report that too.

The same discipline is applied elsewhere on this paper: what is listed against what actually trades on the stock boards, and on Ink, where a venue's entire trading history resolves to seven wallets. Every figure on those pages names the call that produced it, and anyone can repeat the work. That is the whole point of publishing the method beside the number.