Confidential

Microstructure of public token launches — operational framework for EU-based trading

Sector: Digital assets · Year: 2025–2026

Summary

Delivered for a European Union entity. A disciplined programme of launch operations on pump.fun and comparable venues, run as a controlled experiment with complete instrumentation. The work was scoped to measure venue microstructure and operational constraints before any strategy was deployed, and written up as an operating discipline rather than a performance record.

Challenge

Public launch venues are adversarial, fast and largely undocumented. A trader entering one carries uncertainty on three fronts: the venue itself has no published SLA or microstructure documentation, similar-looking launches have wildly different outcomes for reasons that are not captured anywhere, and exit windows close on timescales that defeat manual monitoring. The operational questions — position entry and exit timing, holder cohort dynamics, signal quality of on-chain moves — have no reliable public answer.

Approach

Instrumented launches from genesis: recorded entry prices, exit prices, timing to each exit opportunity, holder growth by time cohort, and the price moves that preceded wash-outs. Each launch instrumented independently so the mechanics are observable without inferring hidden flows.
Built a sample rather than a narrative: ran 64 launches as a cohort, recorded success and failure at the same resolution, and analysed the population rather than the survivors. Tracked which cohorts (by entry timing, size, exit target) succeeded or failed.
Separated venue risk from strategy risk: built monitoring that attributes a loss to a bad venue move versus a bad execution choice, so operational frameworks improve rather than strategies escape blame.
Codified hard limits before starting: position sizing, exit discipline, maximum time-in-position, and a standing rule that any unproven operational change goes to paper simulation before a venue sees capital.

Outcome

A measured body of operational knowledge about launch-venue microstructure covering 64 live launches and 18 simulated scenarios, and a monitoring and risk framework that now travels to any venue before capital is deployed. The framework covers entry timing (best within first 10 blocks), exit timing (60-90s window), position sizing (capped at venue depth), and failure modes (wash-out detection, illiquidity exits). Strategy logic, specific results and capital performance are not published.

Stack

Node, TypeScript, Solana RPC, MongoDB, PM2