Pre-Optimizing ADX to Classify Flat and Trending Markets
Summary
This article proposes testing an indicator in a non-trading pseudo-EA before optimizing a full trading system. Its example evaluates ADX parameters for distinguishing flat from trending markets, using the market’s subsequent price ranges and state durations to score each setup. The optimizer’s custom result combines range separation, average state length, and the number of switches between states.
The first scoring attempt favors implausible configurations, such as flat periods averaging only one bar. The author adds minimum-duration and minimum-switch criteria, then scales the measures with arctangent functions so that very large values do not dominate. In the reported example, narrowing the ADX search reduces the remaining combinations substantially before optimizing the larger EA. These are parameter-search and classification results, not evidence of trading profitability. The scoring thresholds depend on the instrument and sample dates, the initial classification uses ADX level changes, and the author cautions that results may require retuning for other questions or indicators.
Key ideas
- A non-trading pseudo-EA can isolate indicator behavior before optimizing a complete strategy.
- The example classifies market periods as flat or trending using ADX and scores the resulting states.
- Minimum state length and switch count help reject degenerate parameter settings.
- Arctangent scaling balances score components with very different ranges.
- The reported reduction in parameter combinations does not establish trading profitability.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.