What we publish
- Live Kalshi and Gemini market prices on the homepage, fetched from the platforms' public APIs and linked to the markets themselves.
- Aggregate findings from a cohort of profitable Polymarket wallets: where the cohort as a whole is positioned relative to the market price, which strategy types are earning, and how those figures move week to week.
- Fee and expected-value maths using each venue's published fee schedule.
- Evergreen explainers whose worked examples are labelled as hypothetical.
We do not name individual wallets or traders, we do not publish or sell raw data, and nothing here is financial advice. Positions on Polymarket's international, on-chain venue are not the same accounts as Polymarket US customers.
How the smart-money cohort is built
Once a day we read Polymarket's public profit leaderboards for every category over the month and all-time windows, 50 wallets per board. A wallet enters the cohort by appearing on those boards; wallets that appear on more boards rank higher, and we track up to 200. For each one we record its open positions (largest sixty by value), its trade count and its 30-day profit breakdown: market profit, liquidity-provision profit, maker and taker rebates, and reward income.
Raw profit rank is a weak signal on its own. Published research finds that only a minority of top earners are statistically skilled, that most profit accrues to liquidity providers whose edge a follower cannot copy, and that a large share of historical volume looks like wash trading. So before a wallet's positions count toward a published aggregate we drop those whose profit is mostly rebates or liquidity rewards, those with too few resolved markets to judge, and those whose trading pattern looks like volume farming. Aggregates are dollar-weighted and reported with the number of wallets behind them.
Collection began on October 6, 2026. The first aggregate reports appear once there are at least four weeks of snapshots to compare.
How articles are checked
Articles are drafted from a dated fact pack of figures computed from these snapshots and from linked public sources. A separate check reads the finished draft and rejects any number that is not in the fact pack or a cited source. A person reviews and approves every article before it is published, and the publish date and any later update date are shown on the page.
Sources
| Source | What we use it for |
|---|---|
| Polymarket Data API v2 (public) | Leaderboards by category and window, open positions and profit breakdowns of the wallets we track, biggest resolved wins. |
| Polymarket fee documentation | Taker fee rates by category used in the fee calculator. |
| Kalshi fee schedule (PDF) | Taker and maker fee formulas and per-series multipliers used in the fee calculator. |
| Saguillo et al., "Unravelling the Probabilistic Forest: Arbitrage in Prediction Markets" (2025) | Size and concentration of realised arbitrage on Polymarket. |
| Akey, Grégoire, Harvie & Martineau, CEPR discussion paper 21615 | Concentration of profits among liquidity providers. |
| Gómez-Cram, Guo, Kung & Jensen, "Skill vs. luck in prediction markets" | Why profit rank is a weak proxy for skill; persistence of skilled accounts. |
| Sirolly, Ma, Kanoria & Sethi, wash trading on Polymarket | Share of historical volume that looks like wash trading; why volume alone is not a signal. |
Limits
- Leaderboard profit figures are Polymarket's own calculations and can be revised.
- A wallet is not a person: one trader can run several wallets, and some wallets hedge across each other.
- Snapshots are daily, so intraday moves are invisible to us.
- Kalshi does not expose per-trader data, so cohort analysis covers Polymarket only.
Corrections
If a figure looks wrong, tell us with the page and the number. Corrections are noted on the page with the date they were made.