Backtesting basics: rules, data and a fair benchmark

The idea in plain English

A backtest applies specified decision rules to a dataset and calculates outcomes under an execution model. You need to know what information was available at each decision, when a trade could occur and which costs are included. A result without those details is hard to interpret or reproduce.

The distinction that matters

Choose a benchmark appropriate to the question and use the same period and comparable cost assumptions. Check missing data, delisted assets, timezones and corporate or token events where relevant. A clean chart can hide a data problem. Keep the original dataset and configuration with the result.

A practical example

A fictional rule uses the day’s closing price to generate a signal but assumes it also traded at that same price without delay. Depending on the setup, this can use information unavailable at execution. A more realistic next-step execution assumption may materially change the result.

Try this without moving money

• Write the signal timing and earliest possible execution time.
• Record data source, coverage, missing periods and fees.
• Compare with a clearly defined benchmark on the same basis.

A mistake to avoid

Do not accept a result simply because the software produced it deterministically. Deterministic mistakes are still mistakes.

Before you act

Every numerical scenario here is illustrative, not a market quote or forecast. A test can be reproducible and still be wrong about the future. SuperQuant describes a research-only workflow without live order execution. Practice in a paper setting, disclose assumptions, and do not confuse a polished chart with a suitable investment.