Backtest

Structured testing on historical data to estimate edge, limits, and how well the method holds up.

On this page

Who this is for — Anyone with operational rules who wants to know whether they have a real edge before going live: discretionary, semi-systematic, or systematic.

A backtest applies defined rules to a consistent historical sample. It does not predict the future, but it measures how the method reacts to trends, sideways phases, and stress — separating an interesting idea from an operable one.

In plain terms — It is a dress rehearsal on past data: if the method does not hold up there, it will hold up even less in real time.

Bronze prerequisite — Before this lesson: trading-journal, trade-result, trading-mistake. See bronze-path.


What makes a backtest useful

A test is reliable when rules, sample, and costs are realistic:

  • entry and exit rules written without ambiguity;
  • instrument universe consistent with what you will actually trade;
  • costs included (commissions, spread, slippage);
  • stable risk per trade, readable in R;
  • final report with equity, drawdown, and distribution of results.

If these points are missing, the result risks being only a retrospective narrative.

Example — Breakout strategy on an index: without costs it shows +22% per year; with realistic costs it drops to +9% and a worse drawdown. The method remains valid, but it needs different size and expectations.


How to use it without self-deception

  1. Define the rules before opening the data.
  2. Run the test over a long period and across different market phases.
  3. Freeze the method version and move to out-of-sample.
  4. Check robustness with forward-test or live simulation.

Card

  • What it is: retrospective application of operational rules to historical data.
  • When to use it: before going live and every time you change the method.
  • Typical mistake: optimising parameters until the past looks perfect.

Silver path — Module: Validation. Part of silver-path.