Methodology

How we compute PnL

A leaderboard win rate is easy to fake. We count the trades a trader would rather you didn't.

The problem with public leaderboards

Most "90% win rate" profiles are coin-flippers. They redeem winning positions (which shows up as a closed, profitable trade) and quietly abandon losing ones (which never get redeemed, so a naive counter ignores them). The result is a win rate that looks elite and means nothing.

What we actually do

Pull the wallet's full on-chain activity from the Polymarket Data API — every trade, split, merge and redemption, not just the redeemed winners.
PnL = realized + open. Realized PnL comes from resolved and closed positions; open PnL is the current mark-to-market value of positions still live. We show both so a wallet can't hide losses inside unredeemed bags.
Win rate by resolution. We count a market as a loss if the trader's side resolved against them — whether or not they bothered to redeem. Abandoned losers count.
Reconciled to the cent. Our total is checked against Polymarket's own portfolio value endpoint. If the numbers don't line up, we don't publish them.
Freshness-weighted. Recent performance (last ~90 days) is weighted higher, so a wallet that earned its reputation a year ago and has since gone cold can't coast on old glory.

What we can't see — and say so

On-chain data is honest but incomplete. We can't see a trader's off-chain or centralized-exchange activity, and value can be moved between wallets to dress up a profile. When a wallet's flows look like a transfer or wash pattern, we flag it rather than score it as edge. We'd rather show uncertainty than a confident wrong number.