Algorithmic Trading for Prediction Markets

A platform for building, testing and running automated trading bots on prediction markets. You pick a strategy, set its parameters, replay it against real historical tick data, and — if the result justifies it — let it trade. The strategies, the indicators and the risk limits are built in; you supply the judgement.

A market in contracts that settle at a fixed value depending on whether a stated event happens. "Will BTC be above $109,999.99 at 17:12?" is a contract; it pays $1 if it is and nothing if it is not, and it trades between those two numbers until it settles. The price is therefore readable as the market's odds, which is what makes these markets tractable to model.

Kalshi is a US-regulated exchange for prediction market contracts, and it is the venue this platform trades. You need your own Kalshi account and API key to trade live. You do not need one to sign up, to configure a bot, or to run backtests — those work against the archive this box keeps.

No. Every strategy is configured from forms in the dashboard, and every parameter carries an explanation of what it does at its limits. The guide is public and readable before you sign up.

Your configuration is replayed tick by tick over archived market data — up to 30 days of it on the paid tiers — and the run reports net result, win rate, profit factor, drawdown and the trade log behind them. Walk-forward validation splits the window, fits on one half and tests on the other, which is the honest way to find out whether a setting is a strategy or a memory of the data.

Your Kalshi key is sealed in an encrypted vault on the trading engine, which is a separate machine from the one serving this page. It is never sent back to a browser and never appears in a form, a log or an export — the dashboard shows a fingerprint of it so you can tell which key is loaded, and nothing more. You can revoke it at any time. Accounts are protected by a second factor, and live trading is refused outright to an account that has not enabled one.

Position limits, total exposure caps, per-day loss halts and position sizing, all enforced on the trading engine rather than in the page. A browser cannot raise them and a strategy cannot talk its way past them: an order that would breach a limit is refused before it reaches the venue. There is also a stop control on every bot, and a platform-wide one for an operator.

Demo runs the same strategy against the same live market data and places no orders at all — nothing is bought, nothing settles and nothing can be lost. Live sends real orders with your own key. Moving a bot from one to the other is a deliberate step behind its own confirmation, and it is refused unless the account has a second factor and the strategy is one you own.

Yes. This is software for trading a real market with real money, and a backtest that made money is not a promise that a bot will. Markets change, archived data flatters strategies fitted to it, and an automated system can lose faster than a person watching one. Trade only what you can afford to lose, and start in demo.