A Single-Layer Perceptron Strategy Using the Acceleration/Deceleration Oscillator

Summary

This article presents an Expert Advisor that uses readings of the Acceleration/Deceleration oscillator at several lookback offsets as inputs to a single-layer weighted sum. The sign of that score determines whether the system opens a long or short position. For an existing trade, the program can trail a stop loss or reverse direction when the score changes, using a larger position size for the reversal. The example includes configurable weights, stop distance, lot size, and a trade identifier.

The article is also a critique of using neural networks to infer future prices from limited indicator data. It argues for framing the task as a current-state classification problem rather than seeking precise future projections, and favors simple models where market relationships may change over time. It provides implementation details but no backtest, out-of-sample evaluation, or evidence that the strategy is profitable. The oscillator inputs, chosen weights, reversal logic, and risk settings are illustrative, so the code should be read as a modeling example rather than a validated trading system.

Key ideas

• The example scores several lagged oscillator readings with adjustable weights to select a long or short direction.
• The Expert Advisor combines signal reversals with stop-loss trailing and position management.
• The article recommends defining a limited classification task instead of predicting exact future prices.
• No empirical performance evidence is supplied, so the strategy’s profitability remains unestablished.

Tags

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