Plain English in. Real Python out.
Describe the idea the way you’d say it. Premiss writes the strategy as working code you can read.
Premiss turns a trading idea into a real Python backtest on years of crypto market data without code. The code, the trades, and an auditable verdict, in seconds.
Premiss is an AI backtesting tool for research and education. It is a strategy tester for crypto that requires no code. Describe a trading strategy the way you would say it out loud, with no formulas and no code on your side, and the AI builds and runs it as a real Python backtest on years of historical market data. You get a booked, auditable result: every fill closed, every number you can check. A strategy that survives its backtest can continue running as a paper trade on live market data, recording each simulated decision as new prices arrive. Premiss is a backtesting and research tool, not financial advice. Backtested results are hypothetical and do not guarantee future performance.
Describe a strategy. Premiss writes the Python, backtests it on years of market data, and keeps it running as a live bot.
Anyone can have an idea. Almost no one can test it. Premiss is the agent that turns the sentence into a tested strategy.
Everyone has the idea.
Almost no one has the evidence.
Describe the idea the way you’d say it. Premiss writes the strategy as working code you can read.
It reads your idea, builds the strategy, runs it, and reports back with every step visible.
Each fill has a timestamp and a price. No black boxes. You can audit the whole run.
Backtests run on daily candles across full market history while you watch the results come in.
Take a promising backtest and keep it running on live prices, with every simulated decision and trade recorded.
The only thing Premiss needs from you is a sentence.
Running now. 1,462 daily candles checked. The strategy is testing the next historical window.
68.4%completeResult ready. Every entry and exit is logged, so the bounce is more than a good looking chart.
+42.8%returnRunning now. Testing the one week hold after a 10% daily drop, fees included in every fill.
31.2%completeRunning now. Testing each profit target and the fallback exit against the same historical data.
14.6%complete“Hold bitcoin while it’s above its 100-day average. Move to cash when it drops below.”
# A real Python backtest, on real Binance candles. import pandas as pd px = load_candles("BTCUSDT", "1d") # 1,462 daily rows sma = px["close"].rolling(100).mean() # hold while above the 100-day line; cash when below pos = (px["close"] > sma).shift(1).fillna(0) ret = px["close"].pct_change().fillna(0) fees = pos.diff().abs() * 0.001 # 10 bps/switch eq = (1 + pos*ret - fees).cumprod()
+ 3 more trades, each with its own timestamp and final price.
Every run leaves a record
you can go back to.
Choose an example and Premiss writes the code and backtests it.
Premiss is an AI backtesting tool for crypto trading strategies. You describe a trading idea in plain language. It writes the Python, runs the backtest on years of real market data, and returns the code, the trades, and a verified result in seconds.
Premiss is the harness that builds one. You describe the strategy in plain language. Premiss writes the bot's Python, backtests it on years of real data, and can continue running it on live market data.
No. Premiss requires no code. You describe the idea the way you would say it out loud, like “Hold bitcoin while it’s above its 100 day average,” and the AI writes and runs the Python backtest for you.
Premiss backtests against historical crypto market data, including Binance candle data. You can inspect the generated code and the resulting trades, but historical data can’t reproduce every real world cost, delay, or market condition.
Traditional backtesting tools and strategy testers make you write the code or configure indicators by hand. Premiss is an AI backtester: it turns the sentence you typed into real, inspectable Python, so testing an idea takes seconds instead of evenings.
Yes. A strategy that survives its backtest can keep running forward on live market data as a paper trade, recording each simulated decision and trade as new prices arrive.
No. Backtests and paper trades are hypothetical examples based on past market data. They are not financial advice and don’t guarantee future performance. Review the code, assumptions, fees, and trades before relying on any result.
Yes. Premiss is free to start, with daily credits for chatting and backtesting. No card required.