Optimisation addiction: when the strategy tester becomes the hobby
Posted: Mon Oct 19, 2026 3:31 pm
There's a pattern most EA builders recognise: the strategy tester runs all evening, the optimiser tries thousands of parameter combinations, and every morning there's a new "best" set. Weeks pass, nothing goes live, and the parameter sets get more specific each time.
Signs it's become a habit rather than research:
- More time spent optimising than designing new logic.
- The best parameters change every time the date range changes.
- Excitement about a backtest result that you know you wouldn't trust from someone else.
- No forward testing, because "one more run" might find something better.
- Genetic optimisation running on 12 inputs at once.
Ways to break the loop:
- Fix a research budget: e.g. 10 hours on a strategy idea, then a decision to forward test or drop it.
- Limit inputs: no more than three or four optimised parameters.
- Decide in advance what result would count as success, before running anything.
- Forward test on demo for a month before any further optimisation.
The tester is a tool for falsifying ideas, not for finding magic numbers.
Anyone been through this? What finally got you out of the loop?
Signs it's become a habit rather than research:
- More time spent optimising than designing new logic.
- The best parameters change every time the date range changes.
- Excitement about a backtest result that you know you wouldn't trust from someone else.
- No forward testing, because "one more run" might find something better.
- Genetic optimisation running on 12 inputs at once.
Ways to break the loop:
- Fix a research budget: e.g. 10 hours on a strategy idea, then a decision to forward test or drop it.
- Limit inputs: no more than three or four optimised parameters.
- Decide in advance what result would count as success, before running anything.
- Forward test on demo for a month before any further optimisation.
The tester is a tool for falsifying ideas, not for finding magic numbers.
Anyone been through this? What finally got you out of the loop?