Prediction market research you can check.

No. The AI delivers a structured research basis — summary, arguments, probabilities and sources. You always make the decision yourself, and prediction markets involve risk of loss.

No — and you should be wary of anyone who promises that. Prediction markets are hard, and most casual bettors lose by guessing. What we provide is professional research tooling: hundreds of scanned sources per analysis, cited facts, and a probability framework grounded in forecasting science. Outcomes stay uncertain either way.

It follows a fixed framework: identify the key empirical questions, research them across hundreds of web sources including primary documents and the latest developments, apply base rates from comparable past events, weigh the conflicting evidence into a bull and a bear case, and read the resolution rules for hidden clauses — then state its own probability, starting from the market price and moving only as far as the evidence carries it.

All of them. Market Guy works across every Polymarket and Kalshi category — political and election prediction markets, sports betting markets, crypto and Bitcoin markets, and forex, macro and economic events — with the same live data and cited AI research on every market.

A small share of wallets drives most of Polymarket's volume — but a big position is not a good one. Pro+ scores wallets on their edge, the average gap between the outcome and the price they actually paid, across at least 10 resolved markets. It then tells you in one line whether that qualified cohort is positioned against the current price and how strongly, with the evidence behind it. When too few wallets clear the bar, or a single position dominates them, it says exactly that instead of inventing a signal — and the verdict is woven into every AI research report.

For markets that hinge on a price threshold — 'Bitcoin above $150k', say — Pro+ shows the asset's spot price, how far the move still has to go before the market resolves, and how often the asset has covered that distance in a comparable window historically, plus realized volatility and perpetual funding. That historical frequency is a base rate, not a probability: it describes the past, and a change in regime breaks it.

Kalshi is a regulated US prediction exchange that often lists the same events as Polymarket. On Pro+, every market is matched against its closest Kalshi counterpart and shows both venues' prices side by side — a persistent gap can flag an arbitrage opportunity or information asymmetry, and the comparison is woven into every AI research report. Always verify that both markets resolve on the same terms.

Yes. On Pro+, every market is matched against its closest Kalshi counterpart so you can see both venues' prices side by side and spot the gap between them — a persistent gap can flag a prediction market arbitrage opportunity or information asymmetry. Always confirm both markets resolve on exactly the same terms.

We measure it instead of guessing. Our study compares the Polymarket price one week before resolution with what actually happened, across thousands of resolved markets, price range by price range and topic by topic — with the method and the data published at /prediction-market-accuracy. The answer is a historical frequency, not a forecast for the next market.

That the market puts the chance of YES at about 34%. A YES share pays $1 if the market resolves YES and nothing otherwise, so a share bought at 34¢ gains 66¢ when you're right and loses 34¢ when you're not. The guide "How to read prediction market odds" walks through it step by step.

From Polymarket's public APIs, plus Kalshi's public API for the Pro+ cross-market comparison and public exchange data for the Pro+ underlying asset view. We cache prices, volume, orderbooks and trade history server-side and refresh them continuously — you never need a wallet to research.

$3 one time for 7 full days of the plan you pick (Pro or Pro+). Afterwards it converts automatically into that plan — $19/month for Pro, $49/month for Pro+. Cancel anytime.

Pro includes 50 research credits per month on the Sonar model, every topic, a 25-entry watchlist for markets and coins, 90 days of insight history, the Resources and Macro library domains, the full trading guides, AI coin and match scans and the memecoin radar. Pro+ includes 100 credits on the stronger Sonar Reasoning Pro model, unlimited watchlist and history, every library domain with situation briefings, learning paths and the Academy, and three signals Pro doesn't have: the smart-money verdict, underlying asset data for threshold markets and the Kalshi cross-market comparison — each of them woven into every research report.

One credit buys one AI deep-research analysis of a market. Recent analyses are cached, so re-opening a report within a few hours doesn't cost a second credit. Credits reset monthly. During the 7-day trial, analyses are capped at 10 — the full monthly quota unlocks once the trial converts.

Yes. Subscriptions are managed via Stripe and can be cancelled anytime from your account page — you keep access until the end of the billing period.