CAPPR · leaderboard of Polymarket traders
361 traders across nine Polymarket sections — from sports to weather. For each — our Capper Score, the PnL curve, open positions and the cost of copying, computed from transaction receipts. Plus the official summary from the exchange itself. All open and free.
The three groups are the rating's traders split by annual return: the bottom third of the profitable ones, the middle, and the top third. The names describe return only and promise nothing about risk: by our measurements the modest group in fact draws down the hardest — its profit is thinner, and any drawdown eats it whole. The percentages are recomputed from the same table that sits below on this page.
The default order is the CAPPR rating: 70% of the weight goes to the trader’s official profit and 30% to our Capper Score, a composite of statistical reliability, stability, entry-price quality and robustness to one-off wins. Click a row to open a card with the PnL curve, official statistics and open positions.
| # | Trader | Capper Score | Official profit | Return / year | Focus | Open positions | Copytrading |
|---|
Most “top” lists rank by ROI — and a wallet that made 17 000% on a single $480 bet floats to the top. We do not do that: we remove the unprofitable ones and run the rest through five checks.
A bootstrap over 2000 iterations builds an ROI interval. Only those whose lower bound is confidently above zero rank high. An ROI of 8% on 2000 bets is stronger than 80% on fifteen.
We look for the focus on the first 65% of the history and test the edge on the last 35%. The score is high only if the edge held up on “unseen” data.
We recompute the result without the best trade, the top 5 events, the best month. Profit from a single event is not an edge.
The engine reconstructs profit from the trades and reconciles it with the official Polymarket API. The discrepancy on the benchmark: 0.08%.
Traders are selected by AI, not by an editor: every wallet clears the same checks — at least 100 resolved markets, 75 events, 90 active days and $10,000 of turnover — and has to be trading now, not a year ago. A place in the ranking is computed, not granted.