What Kalshi is, what prediction markets are, what the favorite-longshot bias does and doesn't predict, and exactly what Prediction Pilot's AI copilot does (and doesn't do) on your behalf.
Kalshi is a US-regulated prediction market exchange where users can trade binary contracts on real-world events — weather, sports, crypto prices, elections, economic data releases, and more. Each contract pays $1 if the event occurs and $0 if it doesn't. Prices between 1¢ and 99¢ represent the market's implied probability of the outcome.
A prediction market is an exchange where users trade contracts whose payoff depends on the outcome of a future event. The price of a contract reflects the market's collective forecast of the probability that event will happen. Research consistently shows prediction-market prices are competitive with or better than expert forecasts for most domains (Wolfers & Zitzewitz 2004).
Prediction Pilot is an AI copilot for Kalshi. You ask questions in natural language ("is BTC at 92¢ NO a good bet?" or "show me arbitrage gaps over 5¢"), and the AI calls tools that read live Kalshi data, your positions, weather forecasts, Polymarket pricing, and more. It renders the answer as an inline card with the numbers cited. It does not trade for you — every order requires your explicit tap.
Yes. Kalshi is a CFTC-regulated Designated Contract Market (DCM), making it the first federally regulated event-contract exchange in the US. Trading on Kalshi is legal in all 50 states for users 18+.
Kalshi is CFTC-regulated, USD-denominated, and accessible to US residents. Polymarket is a crypto-based prediction market that runs on the Polygon blockchain — historically not legally accessible to US users, though that's evolving. Both list many of the same markets (elections, sports, crypto), which is why cross-platform price comparison matters: a 4-5¢ gap between the two is a recurring source of edge.
The favorite-longshot bias is a market inefficiency in which high-probability contracts trade below their true probability while low-probability ones trade above. On Kalshi, this means 88-95¢ NO contracts win more often than their price implies — historically yielding a 1.5-2.4% systematic edge per trade after fees. Academic literature documents the effect across decades of betting markets (Thaler & Ziemba 1988; Snowberg & Wolfers 2010). See the full strategy guide.
Kalshi charges a per-contract fee of min($0.0175, 7 × p × (1-p)) where p is the contract price as a decimal between 0 and 1. The fee is capped at 1.75¢ per contract and is heaviest near 50¢, where it can reach ~1.75¢. Fees come out of profit on winning trades. There are no deposit, withdrawal, or account fees.
Every Kalshi market has a YES side and a NO side. The YES contract pays $1 if the event happens; the NO contract pays $1 if it doesn't. The two prices always sum to $1 (e.g., YES at 12¢ means NO at 88¢). Buying NO at 88¢ means you risk 88¢ to make 12¢, with an implied 88% probability that the event won't occur.
Yes. You can use the AI to analyze markets, run backtests, scan for arbitrage, and read educational content without ever connecting Kalshi. You only need to connect your Kalshi API key when you want live position tracking, P&L analysis, or one-tap trade proposals.
No. Prediction Pilot is decision support, not automated trading. The AI can propose trades and render them as cards with sizing, EV, and rationale — but every order requires your explicit tap to be sent to Kalshi. The platform never exercises trading discretion on your behalf.
The copilot reads live order books and historical data, your existing positions, external signals (weather forecasts, crypto sentiment, sports records), and competing prices on Polymarket. It synthesizes those into a one-paragraph read with the numbers cited, scored verdicts (A-F grade, EV, sizing), and concrete pass-signals when a trade looks bad. It's a research analyst, not a black-box trading bot.
Every BUY-verdict recommendation is logged and scored against the actual settlement outcome. The copilot exposes its own scoreboard: hit rate vs predicted win rate, by category, with sample size. We don't hide losses — calibration honesty is a feature, and the AI will tell you when its confidence is poorly calibrated against ground truth.
All Kalshi categories: weather (temperature, rain, snow, wind, hurricanes), sports (NBA, NFL, MLB, NHL, soccer, MMA, golf, tennis), crypto (BTC, ETH, SOL, and 15+ others), financials (S&P 500, Nasdaq, Oil, Gold), politics (elections, mentions, policy), economics (Fed rate, CPI, GDP, jobs reports), and culture (Oscars, Emmys, box office).
When the same event trades at different prices on Kalshi and Polymarket, you can buy the cheap side on one platform and sell (or buy the opposite contract) on the other. If the price gap exceeds fees + spread on both sides, the trade locks in profit regardless of the outcome. Prediction Pilot scans both platforms continuously and surfaces gaps of 5¢ or more.
Empirical edge varies widely. The favorite-longshot strategy at 88-95¢ NO has historically returned 1.5-2.4% per trade after fees. Cross-platform arbitrage with 5¢+ gaps returns 3-8% on locked-in trades but is capacity-limited. Weather markets with a forecast cushion ≥ 2°F have shown 70-85% win rates at favorite-band prices. None of these are guaranteed forward returns — past performance does not predict future results.
Describe the strategy in chat ("backtest 88-95¢ NO weather contracts with ≤24h to close over the last 60 days") and the AI calls the run_backtest tool. The result renders as a card with trade count, win rate, total P&L, and per-trade detail. You can iterate by asking "tighten the price band" or "switch to crypto" and a new backtest fires inline.
100,000 free credits on signup — roughly 20 typical questions on Sonnet, ~70 on Haiku. No credit card required. After that, the copilot is metered — $14/month for 2,500,000 credits (roughly 450 typical questions). There's no separate paid tier for features; everything the AI can do is available on the trial.
Prediction Pilot is purely usage-metered at $14/month. There are no Pro/Enterprise tiers that gate features behind plan upgrades — every user gets the same capabilities. Pricing is designed for healthy gross margin (79% at typical use, ~41% at the cap) at the underlying AI's token rates; we don't make money on user inactivity.
Yes. You can export your conversation history, saved strategies, and (if Kalshi is connected) your trade ledger as JSON via the settings panel. We don't lock data behind subscription tiers.
API keys are stored encrypted at rest, scoped per account, and never logged or transmitted to third parties. Kalshi API keys grant the same permissions as your Kalshi account itself, so you should treat them with the same care. You can revoke and rotate keys from Kalshi's dashboard at any time, which immediately invalidates them on Prediction Pilot.
For deeper coverage of any of the topics above, the strategy guides go market-by-market:
Have a question we didn't answer? The AI in-app can probably handle it.
Try Prediction Pilot