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

01 ABSTRACT

Fact: RED-2400 is a public benchmark containing 6,660 algorithmically-rejected trading events from a live Solana DEX filter stack observed from 2026-04-10 to 2026-05-02; 3,333 events link exactly to post-rejection price-and-liquidity sample series, with 169,123 forward-outcome observations and 1,837 lifecycle snapshots covering 1,076 distinct mints. Author's view: The author withdraws Section V comparing how often 'early-death' and other rejected mints reach gone, because 1,235 of 1,236 early-death events were followed within 10.5 minutes by a re-rejection of the same token, so the label marks re-rejection timing, not token death; the author also notes gone is not terminal (167 of 377 mints reaching it later return to alive) and does not denote token death or a rug pull.

02 KEY FINDINGS

1. The dataset contains 6,660 algorithmically-rejected trading events, of which 3,333 (50.05%) link exactly to post-rejection price-and-liquidity sample series.
2. There are 169,123 forward-outcome observations and 1,837 lifecycle snapshots covering 1,076 distinct mints; the deposit contains no outcome-label column.
3. Filter labels are anonymised to filter_1 through filter_8, source-collector identifiers to source_a and source_b; liquidity and 24-hour volume are quantised to the nearest power of two to prevent operational-threshold inference.