Using CatBoost for EURUSD Direction Classification and Trading

Summary

This article introduces CatBoost, a gradient-boosted decision tree library that can process categorical features, and describes its ordered encoding and symmetric tree structure. It compares CatBoost with LightGBM and XGBoost, noting tradeoffs in categorical-data handling, training speed, and dataset scale. The trading example uses EURUSD daily OHLC data together with calendar features. The label indicates whether the next day's close is above its open, and the model is trained to classify that direction before deployment through an ONNX model in MetaTrader.

The article reports that the resulting expert advisor produced 55% profitable trades and $96 in net profit under the stated setup. This is limited evidence: the excerpt does not provide enough detail to assess costs, sample period, robustness, or out-of-sample design, and the random train/test split may not preserve chronological order. The reported trade outcome therefore does not establish that the model will remain profitable in live conditions.

Key ideas

• CatBoost uses gradient-boosted trees and can handle categorical inputs through target-based statistics and ordered boosting.
• The example classifies the next daily EURUSD candle by comparing its future close with its open.
• Calendar fields are treated as categorical features alongside continuous OHLC values.
• The article reports a profitable-trade share and net profit for one expert-advisor setup, without enough detail to establish robustness.

Tags

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