How to optimize without turning the strategy into a moving target
Optimization should refine a stable thesis, not hide a broken one.
- Do not optimize to erase normal drawdown.
- Do not compare strategies that operate under different capital realities.
- Do not keep a bot alive just because the dashboard is interesting.
Start with the result path
Review equity behavior, drawdown timing, and trade clusters before touching parameters. Understand the story first.
Trace the behavior back to the rule set
Ask which market conditions the strategy handled well, which it mishandled, and whether that matches the original design intent.
Change one thing with a reason
Only adjust logic or constraints when you can explain exactly what problem the change is supposed to solve.
Retest before you trust the revision
Every meaningful change resets the burden of proof. A refined rule set still has to earn its place through validation again.