Lorentzian ML Classifier Signals MT4

Limited-time free release: this indicator is free while we collect real trader feedback. Please leave a comment with your requests, issues, settings ideas, or features you want us to consider for the premium version.

Lorentzian ML Classifier Signals MT4 is a closed-bar signal indicator inspired by the high-demand Lorentzian distance classification approach used by many TradingView traders. It compares the current market state against similar historical states using Lorentzian distance, then confirms the result with trend, kernel, and volatility filters.

Use cases

• Find filtered bullish and bearish continuation signals.
• Use confidence values to ignore weak classifications.
• Add machine-learning style context to an existing price action, trend, or risk model.
• Read signal buffers from an EA or strategy tool.

What it shows

• Gold causal kernel trend line on the chart.
• Green arrows for bullish ML signals.
• Red arrows for bearish ML signals.
• Hidden buffers for signal value, confidence, and filter state.

Main features

• Lorentzian-distance nearest-neighbor classification.
• Non-repainting closed-bar signal mode by default.
• Optional trend, kernel-slope, and volatility filters.
• Adjustable training lookback, neighbor count, prediction horizon, and confidence threshold.
• Popup and push notification alerts.
• EA-readable buffers for automation.

Basic use

1. Add the indicator to a liquid symbol and timeframe.
2. Use green arrows only as bullish signal candidates and red arrows only as bearish signal candidates.
3. Prefer signals that agree with your higher-timeframe bias, price action, and risk rules.
4. Raise MinConfidence for fewer, stricter signals or lower it for more examples.

Important inputs

• TrainingLookback: number of historical bars used for classification.
• NearestNeighbors: number of similar historical states used in the vote.
• PredictionBars: forward horizon used to label historical examples.
• MinConfidence: minimum absolute vote confidence needed for a signal.
• TrainingStride: samples every Nth historical bar to keep the classifier responsive.
• UseTrendFilter, UseKernelFilter, UseVolatilityFilter: optional signal filters.
• SignalOnClosedBarsOnly: keeps signals on closed candles by default.

Buffers

• Buffer 0: kernel trend line.
• Buffer 1: bullish arrow price.
• Buffer 2: bearish arrow price.
• Buffer 3: signal value: 1 bullish, -1 bearish, 0 no signal.
• Buffer 4: confidence from 0 to 1.
• Buffer 5: filter state: 1 bullish pass, -1 bearish pass, 0 blocked or neutral.

Risk note

This is a technical analysis tool. It does not guarantee profitable trades. Use it with risk management, higher-timeframe context, and your own confirmation rules.