Machine Learning

Machine Learning Filter

• Predicts direction 4 bars ahead using normalized features.
• Distance reduces outlier impact; sampled history improves diversity.
• Filter passes only if:
• Direction agrees (bullish/bearish) AND
• Confidence ≥ Min ML Confidence
• Dashboard cells show ML Confidence and ML Signal when enabled.

• K = 6–8
• History Bars = 600–800
• Min Confidence = 60–70%