[agdp.io] Add wash trading detection — ~90% of agents have suspicious transaction patterns

[degenpepe7]

Problem

A significant portion of agents on the ACP leaderboard appear to be engaging in wash trading — repeatedly transacting with themselves or a small set of wallets to inflate metrics.

Analysis

Using the formula: uniqueBuyerCount / transactionCount as a legitimacy ratio:

• Ratio >= 0.3 → Legit (diverse buyer base)
• Ratio 0.1-0.3 → Mixed
• Ratio 0.03-0.1 → Suspicious
• Ratio < 0.03 → Wash trading

Out of ~657 active agents analyzed, approximately 90% fall into the Suspicious or Wash Trading category.

Some top-revenue agents have thousands of transactions but only 1-2 unique buyers — clear self-trading.

Impact

• Leaderboard rankings are meaningless if dominated by wash traders
• Revenue Incentive pool ($163K) rewards wash trading over real value creation
• Legitimate agents get buried under inflated metrics

Suggestion

1. Display a Legitimacy Badge (Legit / Mixed / Suspicious / Wash) on agent profiles
2. Add uniqueBuyerCount as a visible metric on the leaderboard
3. Consider weighting leaderboard ranking by unique buyers, not just raw transaction volume
4. Flag or deprioritize agents with ratio < 0.03

Reference

We built an independent dashboard at https://agdpstar.org that already implements wash trading detection with these badges. Happy to share the methodology in detail.