Post-Rejection Follow-up Sampling: Companion Dataset

Companion dataset to Kamat (2026), "Post-Rejection Follow-up Sampling: A Methodology for Counterfactual Outcome Measurement in Algorithmic DEX Trading" (SSRN abstract_id 6607301). Reproduces the paper's headline empirical artefact: 67,000 post-rejection follow-up samples collected by the counterfactual tracking system across eight anonymised filter categories, paired with the 2,997 source rejection events that produced them.

The dataset spans 2026-04-11 through 2026-04-19 (UTC). Filter labels are anonymised to filter_1 through filter_8 in descending order of sample volume; source-collector identifiers are anonymised to source_a and source_b in alphabetical order. Liquidity and 24-hour volume are quantised to the nearest power of two (log2 binning) to preserve heavy-tailed distributional shape while preventing threshold inference. Filter-internal score columns and momentum-derivative price-change columns are removed. SHA-256 checksums for both CSV files are provided in checksums.txt.

Reproducibility cutoff: rejection_outcomes.csv = sampleTs < 2026-04-19T12:18:08.191Z (67,000 rows, matches paper exactly).

A subsequent paper (Kamat 2026c, SSRN 6638259) reports filter-level precision results on a frozen 14-day window using this same data infrastructure; the extended public benchmark (RED-2400, Zenodo DOI 10.5281/zenodo.19989075) covers a wider observation window with full graveyard-tracker coverage.