Holders report.

Each saved holder information snapshot was decoded and compacted. Aggregate metrics were retained alongside selected positions and bounded transaction, transfer, and chart detail. Samples were reduced when required by context limits. All wallets used the same behavioral prompt and 13 questions. The classifications and scores come directly from the model.

Scores use a 0–100 scale. Signals and confidence use 0–1. Bot and human signals are separate judgments, so they do not have to add up to one. Missing values appear as a dash.

We could use these signals to reward real users and explore strategies to counter-trade bots and buy the dips. This report is a saved snapshot, and wallet classifications are JEV’s hypotheses. The complete download includes per-wallet sampling coverage; selected details are not a statistically representative sample.