Original data research into how Polymarket and Kalshi actually trade — on-chain wallet analysis, market pricing studies, and the methodology behind them. Not news coverage, not a rewritten press release: every piece here starts from a dataset we pulled and analyzed ourselves.
Wallet Research
How profitable wallets actually trade, and how rare real, repeatable skill is.
- I Filtered 512 “Profitable” Polymarket Wallets Down to 314 — the filtering methodology behind our wallet tracking: 512 wallets screened down to 314 on consistency of edge, including one wallet that was genuinely skilled and still uncopyable at any realistic position size.
- Inside a Polymarket Esports Wallet That Turned $580K Trading LCK and CS2 — a single wallet’s full position history, read without guessing at intent.
- Smart Money Dossier — 78 tracked wallet profiles, rebuilt from fresh on-chain data every night. Not a one-time study; a live dataset.
Market Pricing & Microstructure
How Polymarket’s prices actually move relative to sharper, faster markets.
- How a Profitable Polymarket Esports Wallet Really Trades a Sharp Book’s Odds — a measured decay curve for how long a pricing gap against a sharp sportsbook survives (full match rate at 0-60 seconds of quote age), plus a population-wide favorite-longshot bias study across 21,792 esports markets.
Trading Strategies
What’s actually been tested, including the strategies that didn’t work.
- Why You Can’t Just Copy Polymarket’s Most Profitable Traders — why most leaderboard wallets can’t be replicated by hand, and the mechanics (speed, position structure, slippage) that explain why.
- Polymarket Betting Strategies That Actually Work.
Research Methods
How we handle data, uncertainty, and the limits of what a dataset can actually tell you — the short version. Full standards on the About page.
- Sources. Polymarket’s Gamma API, Data API, and CLOB API for market and position data; on-chain transaction history for wallet-level work; a sharp sportsbook’s own historical odds, matched by timestamp, where a piece compares pricing speed.
- Anonymization. Individual wallets we profile that aren’t public case studies get no handle and no exact dollar figures — ranges and patterns instead.
- Limitations get stated, not buried. Sample size, timing precision, correlation versus causation — every data piece says what the numbers can’t tell you, next to what they can.
- Out-of-sample where it matters. Where we test a trading rule, parameters are set on one period and tested blind on a later one — not fit and reported on the same data.
Have a dataset question, spot an error, or want to suggest something to dig into? [email protected].