Launch stages: from mint to post-migration
A launch is not one event. It is four, each with its own mechanics, its own visible signals and its own failure patterns. Reading a token page without knowing which stage produced it is guesswork.
The stages are not arbitrary
The four stages this desk uses are not a marketing framework. Each one corresponds to a different program state. At mint, the curve account exists and no swap has touched it. During first buys, the reserve is small enough that individual transactions dominate the quote. During curve fill, aggregate net buying is what matters and individual transactions stop being visible in the price. At migration, the accounting moves to a different program entirely.
Because the underlying state differs, the same observation means different things in different stages. A run of twenty buys in stage two is the entire market. The same twenty buys in stage three barely register. Twenty buys in the first minutes after migration tell you about pool depth rather than about demand.
What each stage tends to break
Stage one breaks on configuration: a metadata field left empty, an image that never resolves, a social link pointing nowhere. These are cheap to fix and expensive to leave. Stage two breaks on distribution: a curve where one or two wallets hold most of the supply produces a chart nobody wants to stand in front of. Stage three breaks on attention decay, which is the most common and least dramatic failure of all. Stage four breaks on depth, because a pool created from a curve reserve starts far thinner than the volume history implies.
The notes below work through the timeline, the graduation moment itself, what happens to volume attribution after migration, and a diagnostic list of the stall patterns that show up again and again.