RED-COHORT-2026-v1: A Catalogue of 1,012 Persistent Wallet Cohorts Detected on the Solana Pump.fun Bonding-Curve Marketplace (June 11-25, 2026)

RED-COHORT-2026-v1 is a public, reproducibility-grade dataset of 1,012 persistentsniper-cohort detections on the Solana pump.fun bonding-curve marketplace, derivedfrom 1,578,333 buyer events across 166,098 token launches observed between2026-06-11 21:43 UTC and 2026-06-25 06:52 UTC (13.38 days elapsed). The release accompanies the working paper: Kamat, A. U. (2026). "Coordinated Sniper Cohorts on Pump.fun: Detection of 1,012 Persistent Wallet Rings and a Contamination-Adjusted Estimate of Coordination- Specific First-Hour Buyer-Flow Lift." Working paper, submitted to Cryptoeconomic Systems (MIT Press). Contents (27 files -- see README.md for the full annotated inventory): sniper_cohorts.jsonl the 1,012-cohort catalogue sniper_cohorts_intra.jsonl.gz raw intra-launch first-buyer observations analyze_sniper_cohorts.py two-stage detection script paper7_v3_psm.py naive PSM script (Section 4.4 estimate) paper7_v3_psm_nocohort_zerofill.py contamination-adjusted PSM script (Section 4.8 headline) psm_results.json naive-PSM result JSON psm_results_nocohort_zerofill.json contamination-adjusted result JSON psm_run.log base-PSM run log psm_run_nocohort_zerofill.log contamination-adjusted run log gen_p7_artifacts.py descriptive-tables and figure generator placebo_bootstrap_v3.py 100-seed activity-matched placebo bootstrap placebo_stability_100seeds.json per-seed placebo distribution fig1_size_distribution_v2.svg cohort size histogram fig2_lorenz_curve_v2.svg cohort-activity concentration fig3_score_vs_launches_v2.svg score vs launches scatter (tier legend) table1_top10_cohorts.csv top-10 cohorts by score table2_size_distribution.csv cohort-size distribution table3_descriptive_stats.csv headline descriptive statistics appendix_a_ablations.csv detection-threshold sensitivity appendix_b_placebo.txt v3-era Design 1 (uniform-random) placebo appendix_b_v2_placebo_cis.txt v3-era Design 2 (activity-matched) placebo with bootstrap CIs appendix_b_tier_stratification.txt v3-era tier-stratified lift breakdown appendix_b_top3_exclusion.txt v3-era top-3-cohort exclusion result DEPOSIT_NOTE.md companion note: v3->v4 supersession, corrected script-to-artefact mapping, environment pin, unreconciled internal-count discrepancies README.md annotated inventory and use notes LICENSE CC-BY-4.0 Methodology and headline results: The detection pipeline runs in two stages. Stage 1 extracts, for each token launch,the ordered list of the first ten buyers within the bonding-curve window. Stage 2builds a co-occurrence graph across launches, retains edges with weight >= 3, runsunion-find to surface connected components, scores each component by10 * (launches hit) + 5 / max(mean first-buyer rank, 1) + sqrt(total SOL committed),and retains components with score above threshold. The full catalogue contains1,012 cohorts comprising 2,965 unique wallets, of which 153 are high-tier(>= 10 launches hit or score >= 100) and 22 are premium (>= 20 launches hit). Under a contamination-adjusted estimator that excludes cohort wallets' own buyerevents from the outcome, 1:1 nearest-neighbour propensity-score matching with a0.2 SD caliper on ten launch-quality covariates yields a first-30-minutenon-cohort buyer-count lift of +16.1% (95% CI [+13.0%, +19.4%], n = 5,419 matchedpairs; worst post-match standardised mean difference 0.077). The correspondingfirst-30-minute non-cohort SOL-inflow lift is +6.3% (95% CI [-0.5%, +15.1%]) andis not statistically distinguishable from zero at 95% confidence. The naive same-universe pooled contrast on this corpus is +130.9% for buyer count;approximately half of that number is arithmetic contamination from counting cohortwallets themselves in the outcome (dropping to +63.9% when cohort events areexcluded), and most of the remainder is absorbed by matching. A parallelactivity-matched placebo estimator, applied across 100 independent seeds, producesa distribution of placebo lifts with median +189.6% (p5-p95 [+166.7%, +217.9%]),well above zero, indicating the activity-matched placebo estimator is itselfbiased; it is retained in the paper's Section 4.5 as a bias diagnostic rather thanas a null baseline. Of 5,419 treated launches, 382 (7.0%) had zero non-cohortbuyers in the first 30 minutes. Reproducibility: The detection script analyze_sniper_cohorts.py accepts a JSONL of buyer events(schema: mint, wallet, slot, blockTime, sol_in, tx_sig, buyer_rank) and producesa cohort catalogue identical in schema to sniper_cohorts.jsonl. All Python scriptsin the bundle use <data-path>/ placeholders for input paths; substitute your localpaths at the top of each script. Both PSM scripts are deterministic under seed=42. The contamination-adjusted scriptpaper7_v3_psm_nocohort_zerofill.py has been re-run on the frozen buyer corpus andreproduces every reported lift, CI, treated/control mean, and covariate SMD toreported precision; the base script paper7_v3_psm.py follows the samedeterministic matching and bootstrap procedure and is expected to reproduceidentically. See DEPOSIT_NOTE.md for the environment pin onplacebo_stability_100seeds.json (RNG-implementation tolerance of 2-5 percentagepoints on distribution statistics is expected across NumPy versions) and for twodocumented internal-count discrepancies (5,411 vs 5,419 strict treated count;77,783 vs 9,788 qualifying pairs) that are internally consistent within their ownconventions. Ethics: All data are public Solana base-58 addresses. No personally identifyinginformation is included. One wallet address in the source corpus included anethnically offensive vanity prefix; in the public release, that prefix has beenredacted to [REDACTED] while the remainder of the address is preserved fortraceability against the underlying on-chain data. Patent disclosure: The detection methodology is the subject of two pending US provisional patentapplications in the author's name at Micro Entity status: No. 64/022,461 (filed2026-03-30) and No. 64/099,108 (filed 2026-06-25). Neither restricts access toor use of the deposited dataset or reproduction scripts, which are released underCC-BY-4.0. Commercial re-implementation may require licensing; [e-mail]. Version history: v1.1.0 (2026-08-01): Added 6 PSM artefacts (paper7_v3_psm.py, paper7_v3_psm_nocohort_zerofill.py, both result JSONs, both run logs) that produce the manuscript Section 4.4 naive and Section 4.8 contamination- adjusted headline. Added DEPOSIT_NOTE.md documenting the v3->v4 supersession, corrected script-to-artefact mapping, environment pin, and unreconciled internal-count discrepancies. Removed redundant v1.0.0 SVGs. Corrected README metadata (13.38-day corpus, v4 paper title, section refs). Catalogue and detection code unchanged. v1.0.1 (2026-07-23): Added v2 figures, 100-seed placebo stability script, updated Fig 3 tier legend. Catalogue unchanged. v1.0.0 (2026-06-27): Initial release.