DashboardMarketsSmart moneyLeaderboardActivityMy portfolio
Insights$POLY CheckerPlatform stats
Live
© 2026 ArrowBow Software Solutions Ltd.·Terms·Privacy
DashboardMarketsSmartLeadersActivityPortfolio

Insights

Models that follow skill, not wealth — every pick paper-tested forward on live markets.

Updated October 2026 · recalibrated continuously

Polymarket insights: skill-rated traders & a self-calibrating model

Traders ranked by skill, not wealth via a Bayesian Edge Score, plus a market model that calibrates itself against every resolved market — so each pick carries a real, backtested win rate.

How the model works

1. Skill, not wealth. Every Polymarket wallet gets an Edge Score — a Bayesian rating that shrinks win rate, ROI and consistency toward the field average by how much the wallet has actually traded, then re-weights it by months-profitable consistency, the one signal that forward-validates. Arbitrage and market-maker bots are detected and zeroed out.

2. Who's positioned where. For each market the model weights the on-chain buy flow by Edge Score to measure how far the skilled money leans past the raw crowd consensus — a self-contained signal that needs no historical price series.

3. Self-calibration.It replays hundreds of recently resolved markets, turning each skilled-vs-crowd lean into a (lean → did-it-win) observation, and fits a calibration table — overall and per category. That table converts the lean into a measured win probability and reports the model's own accuracy and Brier score. It refits every pass as new markets resolve, so the model self-improves.

4. Live picks + a forward record.The fresh table is applied to the hottest live markets, surfacing where skilled money most disagrees with the crowd. Each pick is logged the moment it's made and settled from the same market truth as the live data — a prospective, leak-free track record, reported separately from the backtest.

Want to act on the picks automatically? Learn about Polymarket copy trading →

Polymarket insights — frequently asked

What is a Polymarket Edge Score?
The Edge Score is a Bayesian skill rating for Polymarket traders. Instead of ranking wallets by how much money they have, it shrinks each wallet's win rate, ROI and consistency toward the field average by sample size, so a trader with a long, steady record outranks one wealthy wallet that got lucky once. It is then re-weighted by months-profitable consistency — the one signal that forward-validates (correlation ~0.54 with future consistency) — to surface genuinely skilled traders, not one-hit wonders.
How are the model's pick win rates calculated?
The market model replays hundreds of recently resolved Polymarket markets, measuring for each one how far skilled money (weighted by Edge Score) leaned past the crowd, then whether that lean won. Those outcomes fit a calibration table that turns the lean into a real, measured win probability. The same table is applied live, so every surfaced pick carries a backtested hit rate and the sample of resolved markets behind it — the model refits continuously as new markets resolve and reports its own accuracy and Brier score.
Who are the most skilled Polymarket traders?
The Edge Score board ranks the most skilled Polymarket traders by proven skill rather than wealth — consistent win rate, ROI and months of profitability, with arbitrage and market-maker bots filtered out. The leaderboard refreshes continuously from on-chain trades, so the top of the skill ranking moves as traders perform. Browse the live board on the Insights and leaderboard pages.
Is the Polymarket model's track record real or hypothetical?
It is a forward (prospective) paper record. Every pick the model surfaces is logged the moment it is made and later settled from the same on-chain market truth as the live markets — so the track record reflects decisions made before the outcome was known, not a curve fit after the fact. The backtest accuracy and the forward paper book are reported separately.
Does a high Edge Score guarantee future profit?
No. Skill persists more than luck on Polymarket — months-profitable consistency correlates with future consistency — but prediction markets are adversarial and edges decay as they get crowded. The Edge Score and the model's win rates are probabilistic signals to inform your own research, not guarantees, and nothing here is financial advice.