The first minutes of a launch, read properly
A public feed of new tokens moves faster than anybody can read it, and the fields it shows largest are the fields easiest to produce on purpose. A fast-looking start is not evidence of demand. It is evidence that something happened quickly, which is a much weaker statement.
This desk watches other people launches. It takes a feed row apart field by field, walks the opening window forward minute by minute, and writes down where the reading stops. Every interpretation here is published with the observation that would refute it.
What the opening window contains
- 00:00Creation landsThe mint exists and the curve is initialised. Every figure on the row other than the address and the time is a constant or an assumption.
- 00:20The row becomes visibleEnough activity exists to clear the thresholds on activity-sorted screens, which changes who is watching and therefore what happens next.
- 01:00The first sellsThe record becomes two-sided. This is the first minute that carries information a careful reader can use.
- 05:00A reading that survivesThe sample is finally large enough for a statement about participation to still be true an hour later. Direction remains out of reach.
A description of the general shape of the window. No token, no launch and no screenshot of anybody feed.
What this desk covers
One narrow subject, treated at length: reading a public stream of new launches from the outside, as an observer rather than a participant. Three commitments hold across every page.
Observation before interpretation
Every page separates what the ledger recorded from what an indexer computed and what a creator supplied. Those three are printed as one row and they carry completely different weight.
Where a page interprets something, the interpretation is labelled as one and comes with the evidence that would overturn it.
No predictions and no calls
Nothing here says what a token will do, and no page comments on any named real token. Direction is not something the opening window contains, and pretending otherwise is the source of most of the damage done in it.
The output is narrower on purpose: a defensible statement about participation, dated and checkable.
Numbers only where they can stand up
There are no statistics on this site that we cannot source. No counts of how often anything happens, no ratings, no reviews, no invented feed captures and no example tokens presented as if they were real.
Where arithmetic appears it is labelled as an illustration with round numbers and describes nobody.
Written from the outside
This desk does not run launches, does not operate a token and holds no position it is describing. It is independent of Pump.fun and of every venue named on these pages.
What that buys the reader is the only thing it can: no reason to make a launch look like anything in particular.
The feed
Six pages, ordered by the part of a launch they cover. Each one is a long read rather than a summary, because the shortcuts are what produce the misreadings in the first place.
Watching launches live
The working method: what a feed row is actually made of, which fields come from the chain and which are decoration, and a sequence for reading a new listing without acting on it.
Read 00:00 - 05:00The first five minutes
A minute-by-minute account of what actually exists at second thirty, minute one and minute five, and why the same row reads differently depending on when the feed rendered it.
Read 00:00 - 02:00Reading early buys
How to read the opening trades: order, size, spacing, funding shape, and the four separate reasons a clean-looking distribution can be one person.
Read fill eventWhat a fast fill means
The fill event without the folklore: what the curve mechanically requires, what a fast fill is evidence of, what it is not evidence of, and how migration changes the reading.
Read 05:00 - 60:00When a launch is already over
The quiet signals that a launch has finished before the chart admits it, the difference between a pause and an ending, and what would change each reading.
Read any minuteSix common misreadings
Six specific mistakes, each with the mechanism that produces it, the reason it is convincing, and the check that dissolves it. The bluntest page on this site.
ReadThe activity you are reading was produced by somebody
A meaningful share of the trades visible in an opening window were placed deliberately, by teams paying a vendor for a defined job across a stated set of venues, rather than by a crowd that happened to arrive at once. That is not a rumour and it is not hidden; it is a category with a price list. Looking at how it presents itself, priced and scoped in the open, changes how the first minutes of any launch read.
How the desk works
The method is deliberately unremarkable. Its only unusual feature is that the limits are written down where a reader can hold us to them.
Sourcing
Protocol behaviour is read from published documentation and from the ledger itself. Where a program parameter has changed over time we say so rather than printing a figure that was true once.
Corrections
When something published here turns out to be wrong, the page gets a correction rather than a quiet edit. A desk that silently rewrites its own record cannot ask anybody to check it.
Out of scope
No advice on what to buy, no assessment of any specific token, no legal or tax guidance, and no attempt to identify the people behind any address. All four are outside what a public record can support.
Questions readers send
Twelve of the questions that arrive most often, answered directly and without repeating the articles. Short answers, plainly stated, including the ones where the honest answer is that the record does not say.