Version notes

Post-Rejection Follow-up Sampling: Measuring Outcomes of Rejected Decisions in Algorithmic DEX Trading

Revised description

Note (October 2026): in the deployed data, the tracker sampled each tracked token at a fixed 10-minute cadence and kept one record per token, so the follow-up schedule and horizon described in versions 1 and 2 were not realised. A revised version (October 2026) describes the deployed tracker. The method itself, the deposit and the headline counts are unchanged.

arXiv ↗ SSRN ↗ Manuscript concept DOI ↗ Dataset concept DOI ↗ Code (GitHub) ↗

Outcome-Classified Precision Auditing of Filter Rules in Algorithmic DEX Trading: Evidence from 2,400 Rejection Events

Revised analysis

Note: the posted versions (arXiv, SSRN and Zenodo) are superseded. Their wider headline ratio counted a token's quick disappearance from the price feed as a save; that outcome tier is withdrawn, because in these data follow-up usually ended at the token's next rejection rather than at token death. A rebuilt analysis, which compares the filters' saves and misses with what a fair game (prices with no predictable drift) would produce and replicates the comparison on later data, is in preparation.

arXiv ↗ SSRN ↗ Manuscript concept DOI ↗ Dataset concept DOI ↗

RED-2400: A Public Benchmark of Algorithmically-Rejected Trading Events with Outcome Labels

Documentation updated; data unchanged

Note: the preprint's comparison of tokens that left the price feed soon after rejection with other rejected tokens (Section V) is withdrawn. In these data, follow-up for a rejection event usually ends at the token's next rejection, not at token death, and the lifecycle tracker's "gone" state is not a verified token death. The Zenodo dataset record (version 4.0, September 2026) carries the corrected documentation; the data files are unchanged. A corrected version of the paper is in preparation.

arXiv ↗ SSRN ↗ Zenodo dataset concept DOI ↗ IEEE DataPort DOI ↗ Kaggle ↗ Hugging Face ↗ GitHub reader ↗

Auditing Collector-Generated Graduation Labels on Pump.fun: Measurement Error and Temporal Non-Generalization

Estimates replaced in a later version

Note (28 September 2026, dataset v1.6): the validation failure mainly reflects a calendar-spline term in the frozen model, not evidence that the market or the feature–label relationship changed; the analysis protocol, frozen after earlier analyses of the same data, is a pre-specified re-analysis protocol rather than a pre-registration; and the last day of the validation cohort is outcome-selected. Corrected versions of the paper are in preparation.

Note (September 2026): the current arXiv version (v4) withdraws the survival-analysis estimates of the earlier versions, including the pooled 0.198 percent graduation rate, the "3.18x decline" built on it, the Telegram "lift" figures and the wallet-intent narrative, because each treated the collector's label as the platform's graduation event. The earlier arXiv versions (v1 to v3) and the SSRN posting still show those claims and should not be relied upon.

arXiv ↗ SSRN ↗ Zenodo dataset concept DOI ↗

Counterfactual Rejection Filter Ablation: An Empirical Application of Post-Rejection Follow-up Sampling to Solana Memecoin Trading

Revised analysis in preparation

Note: the per-filter summaries in the July 2026 deposit attribute each token to its first rejection reason and are superseded. A revised analysis at the level of individual rejection events is in preparation.

Zenodo dataset concept DOI ↗

A Multi-Venue Solana/DeFi Microstructure Data Corpus: The RED-2400 Family v2

Documentation updated; data unchanged

Note (September 2026): version 1.1 of each dataset (28 September 2026) corrects the documentation; the data files are unchanged. Descriptions in the posted paper that differ from the version 1.1 records (for example, liquidation counts, the number of chains in the utilization panel, and DEX "venue" labels, which are the first leg of an aggregator route) should be read as corrected there.

SSRN ↗ RED-ORACLE Zenodo ↗ RED-ARB Zenodo ↗ RED-LIQ Zenodo ↗ RED-BASIS Zenodo ↗ RED-BRIDGE Zenodo ↗

RED-2400 Replication Toolkit: A Python Package for Reproducible Filter-Precision Auditing

Software updated

Release v1.0.0 pooled outcomes by token, so its per-filter tables and save-to-miss ratios do not measure per-event outcomes and should not be cited as findings; it is kept only to reproduce the posted filter-audit versions. Release v2 attaches outcomes to each rejection event.

GitHub ↗

Dev-Wallet Silence Is Not a Rug Predictor on Solana Pump.fun: Empirical Evidence from a Positive-Class and Control-Panel Study

Further revision planned

Corrected v1.3 (26 August 2026). Small-sample study; the dev-wallet and rug labels are the data collector's own classifications, not verified on-chain facts. A further revision is planned at the journal stage.

Zenodo concept DOI (latest) ↗

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

Withdrawn by the author

Note (October 2026): the buyer records used in this analysis and in the RED-COHORT-2026 record include sell transactions and miss many buys, so all numerical results, including the cohort count in the title and every buyer-flow estimate, are withdrawn. No effect of early-buyer groups is claimed. Any corrected analysis will be posted as a new version.

arXiv ↗ Results and code record ↗

Preregistration: Hour-Aware Prospective Deployment Test (OSF)

Registration replaced

The registration of 5 October 2026 (osf.io/yvb3m) replaces a September 2026 registration, which was withdrawn before any data were collected; the new registration is a redacted version.

OSF ↗