01 Research Question
We seek to evaluate how verifiable on-chain behavior can be used to assess the risk of "smart-money consensus signals" rather than purely chasing leaderboard PnL. Public data lets us track real liquidity shifts and portfolio changes, but no single address guarantees future price trajectory. We need reproducible, verifiable methods to map out the underlying risk structure.
"A single profitable trade is not a reproducible strategy."
02 Data & Methods
We construct smart-money samples from public transaction ledgers, normalizing cost bases, filtering cluster addresses, and removing anomalous noise to reflect actual capital intent.
| Dimension | Processing | Objective |
|---|---|---|
| Realized PnL | Computed from exit transactions | Eliminate unrealized paper gains |
| Cost Basis | Executed price plus network fees | Prevent double counting |
| Clustering | Address clustering via interaction graphs | Prevent multi-account distortion |
| Wash Filter | Flag and exclude circular transfers | Minimize signal contamination |
03 Consensus and Exit
Taking PEPE as a reference case, we observe the interaction between consensus wallet holdings and centralized exchange net inflows from July to October 2026.