Multi Kernel Regression MT4 & MT5

Products:

Introduction

This tool takes raw, noisy price data and transforms it into smooth, easy-to-read trend lines that make market direction much clearer.

Whether you’re a day trader, swing trader, or long-term position trader, this indicator can adapt to your style thanks to flexible kernel selection, adjustable bandwidth, and two different calculation modes.

What is Kernel Regression?

Imagine you’re looking at a messy scatter of price points and you want to draw a smooth curve that shows the overall direction — without just connecting every single point. That’s exactly what kernel regression does.

Instead of fitting one fixed mathematical line (like linear regression), kernel regression builds a smooth value at each bar by giving more importance to nearby prices and less importance to distant ones. This “importance” is controlled by something called the kernel function.

Different kernels weight prices in different ways, which means you get different-looking trend lines from the same price data.

A simple way to think about it: imagine you’re standing on one candle. You look at nearby candles and they influence your decision a lot. Candles far away still matter, but much less. The kernel decides how fast that influence fades with distance.

Key Features Overview

• 17 Kernel Types: Gaussian, Logistic, Triangular, Cosine, Linear, and 12 others for different market behaviors
• Dual Calculation Modes: Standard mode for maximum accuracy, Recalculate mode for smoothing out the noise
• Volatility Bands: Optional standard deviation channels that expand and contract with volatility
• Band Crossing Signals: Automatic arrow alerts when price crosses the bands
• Flexible Bandwidth Control: Adjust from very responsive to very smooth
• Performance Optimization: Control update frequency and how much history is recalculated
• Platform Agnostic: Works the same on MT4 and MT5

How to Use the Multi Kernel Regression Indicator

Understanding the Visual Components

• Main Regression Line: The smoothed trend line based on your kernel and bandwidth.
• Upper and Lower Bands (optional): One standard deviation away from the main line. They widen when volatility increases and narrow when volatility decreases.
• Arrows: Appear when price crosses the bands. Up arrow = price crosses above the lower band. Down arrow = price crosses below the upper band.

Quick Settings Cheat Sheet

For Scalping (1–15 min)

• Kernel: Gaussian or Triangular
• Bandwidth: 20–40
• Recalculate Mode: Off
• Refresh After Ticks: 1–2

For Swing Trading (4H–Daily)

• Kernel: Gaussian or Cosine
• Bandwidth: 50–100
• Recalculate Mode: On (Multiplier 2.0)
• Refresh After Ticks: 0

For Trend Following (Daily–Weekly)

• Kernel: Gaussian
• Bandwidth: 100–200
• Recalculate Mode: On
• Refresh After Ticks: 0

For Support & Resistance

• Kernel: Triangular or Linear
• Bandwidth: 30–60
• Enable Deviation: Yes
• Use bands as dynamic S/R

Trading Approaches

• Trend Confirmation: If price stays above the line, trend is up. Below it, trend is down.
• Mean Reversion: Trade the bands. Upper band = potential overbought. Lower band = potential oversold.
• Direction Changes: When the slope flattens or turns, trend change may be coming.
• Signal Generation: Arrows show band crossings. Great in trends, weaker in chop.

Understanding the Calculations

The Core Concept

At its heart, the indicator calculates:

Regression Value = (Sum of [Price × Weight]) / (Sum of Weights)

Unlike a moving average, not all prices are weighted equally. Nearby prices matter more.

How Weights Work

• Very close bars = strong influence
• Medium distance bars = medium influence
• Far away bars = very small influence

This behavior is controlled by:

• Kernel Function: Shape of the weighting curve
• Bandwidth: How far influence extends

Bandwidth Explained

• Low (20–40): Very responsive, more noise
• Medium (50–100): Balanced
• High (150+): Very smooth, slower to react

The 17 Kernel Types Explained

Most Common

• Gaussian: Smooth, natural, best all-rounder
• Triangular: Sharper and more responsive
• Logistic: Similar to Gaussian but turns faster
• Cosine: Good for cyclic behavior

Specialized

• Linear, Uniform, Epanechnikov, Hyperbolic, Quadratic (Biweight)

Advanced

• Quartic, Triweight, Tricube, Sinc, Lanczos, Welch

Tip: Just use Gaussian first. Bandwidth matters more than kernel choice.

Settings Explained in Detail

Core Parameters

• Kernel Select: Triangular, Gaussian, Epanechnikov, Logistic, Log-Logistic, Cosine, Sinc, Laplace, Quartic, Parabolic, Exponential, Silverman, Cauchy, Tent, Wave, Power, Morters
• Bandwidth (5–500)
• Price Source

Volatility Bands

• Enable Std Dev
• Deviation Multiplier

Calculation Modes

Standard Mode (Recalculate OFF): No repainting. Each bar calculated once. Arrows visible.

Recalculate Mode (ON): Bars recalculated based on Bandwidth x Multiplier. Arrows hidden.

If you keep StartBar 0, then even with Recalculate false, it might still repaint the arrow while bar is still forming.

So default StartBar (shift bar) is starting from 1 when current bar is fully closed.

Signal Controls

• Show Arrows
• Enable Std Dev

Standard Mode vs Recalculate Mode

Standard Mode

• Best for scalping and signal trading - sideways market. Be careful during trending market and don't trade against the trend.

Recalculate Mode

• Smoothing of trend lines and deviation bands
• Best for trend detection
• Arrows hidden (because they might repaint)
• Best for multi-indicator setups and to remove market noise

Arrows only appear in Standard Mode because Recalculate Mode can retroactively change previous bars, which would cause arrows to appear and disappear.

• Up Arrow: Price crossed below and then back above lower band
• Down Arrow: Price crossed above and then back below upper band

Practical Examples

Swing Trading

• Gaussian, Bandwidth 75, Recalculate ON, Deviation 1.5

Scalping

• Triangular, Bandwidth 35 or lower, Recalculate OFF, Arrows ON

Support & Resistance

• Gaussian, Bandwidth 100, Deviation 2.0

Conclusion

The Multi Kernel Regression Indicator is a powerful trend smoothing and structure tool.

Start simple: Gaussian kernel, bandwidth around 50. Then adjust from there.

Use it with price action, structure, and common sense. It’s a confirmation tool, not a magic button.

Whether you scalp or swing trade, there’s a kernel and bandwidth combo that fits your style.

Happy trading!