Using Random Forests and Rattle to Classify Currency Trends
Summary
This article outlines a workflow for using Rattle, a graphical interface to R statistical tools, to explore classification models for predicting whether currency prices will rise or fall. It distinguishes classification targets from price forecasting: the target is a future long or short label, while predictors include prices, time categories, and technical-indicator values. Historical ZigZag turning points define the target labels, which are shifted forward by one bar for a one-step-ahead prediction.
The example uses hourly data for six currency pairs over a stated historical period, creating a dataset with 88 predictors. Rattle is presented as a way to compare candidate models, inspect predictor importance, and reduce inputs while monitoring model quality; a trained model could then be connected to MetaTrader 4. The excerpt does not give complete model results or establish live trading profitability. ZigZag repaints at the chart’s right edge, so it is used to label historical targets rather than as a live predictor, and the choice of target and predictors remains a central modeling judgment.
Key ideas
- Trend prediction is framed as a classification task with long and short target classes.
- Historical ZigZag turns supply labels, shifted forward to represent a future bar.
- Predictors include currency prices, time categories, and technical indicators across six pairs.
- Rattle supports model comparison and predictor-importance review before implementation.
- The described workflow is exploratory and does not demonstrate live profitability.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.