Engineering Neural Network Inputs from Indicator Distances, Accumulations, and Slopes

Summary

This article proposes transforming technical indicators into features that may be more informative for a perceptron than raw prices or indicator levels. It defines three approaches: distances between indicators or between current and historical values, accumulated indicator differences to represent persistent directional movement, and slope angles over a chosen history to describe impulse or fading activity. Examples use moving averages, MACD, and CCI, with code illustrating current and lagged differences as inputs to a weighted perceptron.

The author says forward testing showed the approach raising the balance confidently over its first six months, but the provided text has no test figures or detailed performance analysis. It presents illustrative EA implementations rather than a controlled comparison against raw inputs or alternative feature designs. The suggested features therefore remain experimental: their usefulness may depend on the instrument, timeframe, scaling, and validation method, and the article does not establish general profitability or robustness.

Key ideas

• Indicator differences can relate price-derived features to a reference value or historical observation.
• Accumulated differences are presented as a way to distinguish sustained movement from choppy consolidation.
• Slope angles over a fixed candle history can describe the strength and direction of an indicator's movement.
• Examples combine moving-average, MACD, and CCI features as inputs to a perceptron.
• The reported forward-test claim is not accompanied by detailed performance figures in the text.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.