Each group below enforces a published floor — a minimum resolved sample plus a profit threshold — and lists every wallet that clears it, ranked by realized profit. The floor is stated on every group, so you can see why a wallet qualified before you follow it.
Standing cohorts
Fixed floors with no time window. Membership only changes when a wallet's rollup crosses a threshold.
OpenStanding
Profit Leaders
Wallets up more than $25,000 on our own ledger — profit reconstructed fill by fill from chain data — over a real sample of resolved bets.
≥ 50 resolved bets≥ $25,000 PnLbots and market makers excluded
Defined but not published: the rollup they read cannot support their claim yet. None carry members or link anywhere until it is rebuilt.
Category rollup is stale (last written 2026-05-07) and its win/loss counters do not reconcile. Pending rebuild.
Sports Sharps
Pending data
Politics Sharps
Pending data
Crypto Sharps
Pending data
Macro Sharps
Pending data
Culture Sharps
Pending data
Breaking-News Sharps
Pending data
What these groups are, and what they are not
How do you decide which Polymarket wallets are worth following?
Each group enforces a published floor — a minimum number of resolved bets plus a profit threshold — and every wallet that clears it is listed, ranked by realized profit. Membership is recomputed on read, so a wallet that stops clearing the floor drops out on its own.
Is a high win rate enough to call a Polymarket wallet good?
No. A win rate over a handful of bets is noise, and betting heavy favorites produces a high win rate with no edge. These groups rank on realized profit and show a sample-size-shrunk win rate instead of the raw one — which is deliberately not displayed — so a 100% record over 20 bets never outranks a profitable wallet with a deep sample.
Can I pull these wallet groups from an API?
Yes. GET /api/v1/cohorts lists every group with the floor it enforces, and GET /api/v1/cohorts/{slug} returns the members with their resolved bets, profit and win rate. The same data is exposed to AI agents through the CrowdIntel MCP server as list_cohorts and get_cohort.
How current are the figures?
They come from pre-aggregated wallet rollups over every on-chain Polymarket fill since 2020, not from a live order book, so each group reports the timestamp its underlying data was last written. Treat the figures as of that timestamp, never as real-time.
