> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tqx.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Parameter optimization

> Search a defined parameter space and evaluate the resulting strategy versions.

Parameter optimization tests several values for parameters exposed by the strategy. It helps you explore sensitivity; it does not remove the need for out-of-sample validation or risk review.

## How optimization works

<Steps>
  <Step title="Choose the strategy version">
    Start from the version whose code and backtest settings you understand.
  </Step>

  <Step title="Define the search space">
    Ask Qube to identify tunable parameters or specify the candidate values yourself. Keep the grid small enough to inspect and explain.
  </Step>

  <Step title="Run the combinations">
    Qube runs the selected combinations and tracks each result in the optimization task. Failed combinations remain visible with their failure reason.
  </Step>

  <Step title="Review the best result and sensitivity">
    Compare the objective metric, equity curves, trade count, and drawdown. Check whether nearby parameter values behave similarly.
  </Step>

  <Step title="Apply and verify">
    Applying the selected parameters saves a new strategy version. Run a fresh backtest and compare it with the prior version before treating the change as an improvement.
  </Step>
</Steps>

## What to review

* The objective used to rank combinations, such as Sharpe ratio.
* Total return, annualized return, maximum drawdown, and trade count.
* Whether the best result is an isolated peak or part of a stable region.
* Whether the selected period and costs are realistic.
* Whether the result survives a different or later validation period.

<Warning>
  Do not select a parameter set only because it has the highest historical return. A more stable result with lower drawdown may be a better research candidate.
</Warning>