Sampler Indicator Signals for Neural Network Training
Summary
The Sampler indicator is described as a way to derive training targets from future price changes. One output is an analog value normalized between negative and positive one, representing the balance of upward and downward price movement over a specified forward window. A second output converts the result into discrete buy, wait, or sell labels. The document gives two ways to form those labels: compare the analog value with a threshold, or assign a signal based on take-profit and stop-loss conditions.
An accompanying expert advisor is intended to check indicator values using data written to a file. The description does not provide formulas for the take-profit and stop-loss method, parameter-selection advice, or validation results. Because the target uses future bars, it is suitable as a retrospective label for training or evaluation, not as information available when making a live decision; the document does not explain safeguards against look-ahead leakage in model workflows.
Key ideas
- The analog output summarizes normalized price movement over a forward bar window.
- The discrete output labels conditions as buy, wait, or sell.
- Discrete labels can be based on an analog threshold or take-profit and stop-loss conditions.
- An expert advisor checks indicator outputs using data saved to a file.
- Future-based labels require care to prevent look-ahead leakage in model evaluation.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.