Gaussian Naïve Bayes for Classifying Short-Term Forex Direction

Summary

The article applies Gaussian Naïve Bayes to classify whether a bar closes above or below its open, using Bulls Power, Bears Power, RSI, tick volume, and Money Flow Index as features. It explains preparing a labeled matrix, splitting observations into training and test sets with a repeatable shuffle seed, fitting the classifier, and reviewing confusion matrices and classification metrics. It also outlines Bayes' rule, class prior probabilities, and the model's assumption that features are conditionally independent, despite the observed correlations among several indicators.

The reported EURUSD example gives a 58% training-set accuracy and shows uneven class performance; these figures are not evidence of out-of-sample profitability. The article proceeds to test data and describes using the model in an Expert Advisor, while cautioning that strategy logic and tester performance need careful evaluation before live use. Its small example, bar-direction label, and correlated inputs limit what can be inferred about predictive value.

Key ideas

• The example labels each observation by whether its close is above its open and uses five technical inputs to predict that class.
• Gaussian Naïve Bayes combines class priors with feature likelihoods under a conditional-independence assumption.
• The article checks correlations among inputs, which are substantial for several of the chosen indicators.
• A repeatable shuffled train-test split and confusion-matrix metrics help inspect classifier behavior.
• The reported training accuracy does not establish out-of-sample trading performance or profitability.

Tags

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