pick_best_grid_search_result#

pick_best_grid_search_result(results, key=<function <lambda>>, highest=True)[source]#

Pick the best combination in the results based on one metric.

Use trading metrics or performance metrics for the selection.

Example:

sample = pick_best_grid_search_result(
    results,
    key=lambda r: r.metrics.loc["Max Drawdown"][0])
    assert sample is not None
Parameters:
  • result – Output from perform_grid_search()

  • key (Callable) –

    Lambda function to extract the value to compare from the data.

    If not given use cumulative return.

  • highest – If true pick the highest value, otherwise lowest.

  • results (List[GridSearchResult]) –

Returns:

The grid search result with the matching parameters or None if not found

Returns:

The grid search result with the matching parameters or None if not found

Return type:

Optional[GridSearchResult]