Skip to main content
Coinversa indexes every active Hyperliquid wallet and classifies it into two orthogonal tier systems:
  • PnL tier — by total profitability across the wallet’s lifetime
  • Size tier — by total volume traded
A wallet always has exactly one PnL tier and one size tier. Together they form a 2D map of trader behavior — e.g. “Sharps + Middleweights” is a different cohort from “Sharps + Featherweights”, even though both are profitable. Each tier has a canonical slug (use this in requests) and a legacy slug. Dual-vocabulary input — accepting both slugs on path parameters, query parameters, and tool arguments — is rolling out now; until it reaches your endpoint (MCP connectors may also cache tool schemas that still list only the legacy slugs), fall back to the legacy slug if a new one is rejected. Responses currently return the legacy slugs (e.g. pnlTier still comes back as money_printer) — match on either when reading responses.

​
PnL tiers

Ordered from best to worst:

​
Size tiers

Ordered from largest to smallest:

​
How to query cohorts

Use the cohort tools when you want behavior at the group level rather than per wallet: Cohorts are addressed by tierType (pnl or size) and tier (a slug from the tables above — new slugs are canonical; legacy slugs always work, and are the safe fallback until the dual-vocabulary rollout reaches your endpoint):
If a new slug is rejected (for example, an MCP tool schema that still lists only legacy slugs), send the legacy equivalent — smart_money for sharps, whale for middleweights — from the tables above.

​
When to use cohorts vs individual wallets

  • Cohorts when you want a signal — “are the Middleweights loading up on BTC?” / “are the Sharps bearish on ETH?”.
  • Builder cohorts when you want the user mix of a builder code — the builder endpoints label wallets with their lifetime tiers (pnlTier / sizeTier, legacy slugs), which can differ from the 30d-rolling tiers the cohorts-recent endpoints classify by. See Builder codes & attribution.
  • Individual wallets when you want a specific story — “what is 0xabc... doing?” / “how is this trader I follow performing?”.
Most agent workflows benefit from cohort-first analysis: it’s signal-rich, and you can drill into specific wallets with pulse_token_leaderboard (top traders for a coin) when you want names.