Encoding Moving-Average Geometry for Perceptron Trading Signals
Summary
This article explores representing price behavior as geometric features for a perceptron rather than feeding it raw prices or oscillator values. The proposed inputs use two simple moving averages, with periods of one and twenty-four, and describe their spacing and changes across selected closed candles. Example feature shapes include line segments, envelopes, quadrilaterals, and combined structures. A related representation measures slope using price change per bar, avoiding chart-scale-dependent angle calculations. Values are expressed in points to keep inputs within a more consistent range.
The author tests these features in a counter-trend strategy: moving-average ordering establishes the trade side, and a weighted perceptron score gates entries. The article reports optimization settings and describes a result and a forward test, but leaves the forward test as further work and acknowledges unresolved issues in managing optimized parameters. The evidence is an individual experiment, not a controlled comparison. The fixed stop and target, selected instrument and timeframe, and extensive weight optimization limit how broadly its reported performance can be generalized.
Key ideas
- Moving-average distances across past closed candles can encode price structure as perceptron inputs.
- The examples construct geometric features from two moving averages, including spacing, segment changes, and combined shapes.
- Slope features use price change per bar to reduce dependence on chart scaling.
- A counter-trend entry rule combines moving-average ordering with a weighted perceptron score.
- The reported experiment relies on optimized weights and has unresolved forward-testing and parameter-management limitations.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.