Reading the result
The instinct is that a profit leaderboard is full of market makers and lucky runs that cannot be followed. A flat copy tests that directly. 79 of 94 scored Kalshi traders and 81 of 94 Polymarket traders finished the window in profit at a flat $100 per entry, and 74 and 61 are still in profit after 4 points come off every return for slippage and fees.
Kalshi leaders won a median 74.9% of settled entries and returned a median +18.3% on stake. Polymarket leaders won a median 64.4% of settled entries and returned a median +8.1% on stake. Win rate does not set the return by itself. A trader who wins often at 80 cents and up earns little on each win, and a trader who wins less often at lower prices earns more on each one.
The top ten on the Kalshi board produced 8 profitable copies and the bottom ten produced 6. The top ten on the Polymarket board produced 8 profitable copies and the bottom ten produced 8. Across both boards a group of ten produced between 6 and 10 profitable copies. The number one trader on a profit board can be a losing copy when the profit came from sizing a few positions enormously rather than from a per-entry edge, which is exactly what a flat stake removes. The variables that do predict a good copy are the ones the leaderboard hides: win rate against the price paid, the depth of the settled record, and whether the profit came from many entries or a handful. Every name above links to a public record that shows all three, and the backtester runs the same simulation on any rule.
Population and method
- Board. The top 100 traders per venue by observed net profit on Rivo's recorded trades, ranked at 2026-09-23T03:49:50.818Z. The same ranking as the public traders page.
- Window. Recorded entries from 2026-06-25T03:49:50.818Z to 2026-09-23T03:49:50.818Z, the trailing 90 days at export.
- Eligibility. buy_yes and buy_no entries with a recorded price. Sells are never scored. Entries in markets still open at export are counted but not scored.
- Stake. $100 per entry, filled at the recorded price, held to settlement. PnL is 100 / price − 100 on a win and −100 on a loss.
- Scoring floor. A trader needs at least 30 settled entries to enter any share, median or count. Thinner records are listed and downloadable, not scored.
- Cost case. 4 return points subtracted from every scored trader, a flat stand-in for slippage and fees.
- Not modelled. Entry delay, partial fills, depth, position limits, capital lock-up, and correlation between a trader's entries in the same market.
The full definitions, including the settlement rule and the payoff arithmetic, are on the methodology page.