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

Battle Count: 66

Highly relevant to quantitative trading in the DEX/cryptocurrency domain. The paper directly addresses filter precision measurement, counterfactual evaluation of rejected trading candidates, and the validation of outcome-classification rules used in algorithmic trading systems. The save-to-miss ratio analysis and the finding that the 'early-death' classification tier does not carry incremental signal beyond rejection itself are directly actionable for trading system design. However, it is a dataset/benchmark paper rather than a novel trading strategy or model paper, and its scope is limited to Solana DEX trading.

Low implementation complexity. This is primarily a dataset deposit paper with a validation analysis. The reproduction involves reading CSV files, performing joins on the mint column, and computing value_counts. No model training or complex algorithmic implementation is required. The five-tier classification rule is described in a companion methodology paper. Standard data manipulation in pandas or Node.js suffices.

Highly reproducible. The dataset is publicly available on Zenodo under CC-BY-4.0 license with SHA-256 checksums for all three CSV files. Reproduction commands are bundled with the deposit README. Row counts, per-filter shares, graveyard to-state vocabulary, and the early-death validation join all reproduce in standard pandas or Node.js environments. The methodology paper's companion deposit (DOI 10.5281/zenodo.19987695) is also publicly available with an audit script.