What 'AI backtesting' actually means
AI backtesting uses a model to help write the strategy logic and prepare a simulation on historical data. Depending on the tool, it can also load data, execute the code, and explain the output.
The result describes how the implemented rule behaved under the test's assumptions. It does not show that the model can predict prices or that the strategy will perform similarly in a live account.
How it works, step by step
A backtest workflow has four stages:
- Read your intent: parse 'hold bitcoin above its 100-day average' into precise rules.
- Generate the code: write the actual backtest logic for those rules.
- Run it on real data: execute against years of real candles, stepping through time without lookahead.
- Show the result: return the trades, the equity curve, and summary metrics - ideally with the code visible.
Keeping AI backtesting honest
Check realistic costs, data coverage, and the timing of each decision. Reserve a period that was not used to choose the rules, and vary the inputs to see whether the result depends on one narrow setting.
Inspecting the generated code and individual trades can expose mistakes that summary metrics miss. For example, verify that a trade triggered by a daily close is not filled at a price from earlier that day.
How Premiss does it
Premiss uses your description to generate a Python strategy and run it against historical market data. The workspace includes the code, trades, and performance metrics, so you can review what was tested and revise the idea with that evidence in view.