Coordinated Sniper Cohorts on Pump.fun: Detection of 1,012 Persistent Wallet Rings and the Limits of Naive Causal Inference for First-Hour Buyer Flow

Arati Uday Kamat

Papers from arXiv.org

Abstract: We study coordinated buyer behavior on the Solana pump.fun bonding-curve marketplace using 1,578,333 buyer observations from 166,098 token launches between June 12 and June 26, 2026. A two-stage detection pipeline - intra-launch first-buyer-window extraction followed by cross-launch persistent-cohort surfacing via union-find on co-occurrence graphs - identifies 1,012 persistent wallet cohorts (2 to 12 wallets each, drawn from 2,965 unique addresses) that systematically co-fire as early buyers across multiple launches. 153 qualify as high-tier (>= 10 launches or score >= 100); 22 hit >= 20 launches. The top cohort (9 wallets) appears among the first 10 buyers of 42 distinct launches over 11 days at average first-buyer rank 2.29. Under a 3:1 random-matched design, cohort-touched launches show a +132.3% lift in first-30-minute buyer count (95% CI [+127.0, +137.4], n_treated = 5,411) and +136.5% lift in SOL inflow (95% CI [+120.9, +152.2]). An activity-matched placebo drawing wallets matched on per-wallet launch-count yields a LARGER buyer-flow lift of +216.3% (95% CI [+183.8, +255.2]). The real-cohort estimate falls well below the placebo CI lower bound, refuting a strong cohort-specific causal interpretation. We interpret the association as evidence of selection: both real cohorts and activity-matched placebos disproportionately appear in the buyer queues of launches attracting elevated buyer flow for reasons orthogonal to coordination. Identifying a cohort-specific causal effect requires propensity-score matching on launch-quality covariates. We release the full cohort catalogue, detection code, and robustness artefacts as RED-COHORT-2026-v1 under CC-BY-4.0.

Date: 2026-07
References: Add references at CitEc
Citations:

Downloads: (external link)
https://arxiv.org/pdf/2607.02795 Latest version (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:arx:papers:2607.02795

Access Statistics for this paper

More papers in Papers from arXiv.org
Bibliographic data for series maintained by arXiv administrators ().