Understanding Kernel Settings

Advanced parameters that control how kernel regression estimates the trend and adapts to market fluctuations. Kernel functions estimate the underlying trend by weighting past price data non-linearly, offering more flexibility than standard moving averages.

Interactive Settings Explorer

Below is an interactive exploration of the Kernel Settings interface. Click any setting to instantly view its detailed documentation, default value, and purpose:

Lorentzian Classification [2.06 - Beta]

⚙️ KERNEL SETTINGS

Trade with Kernel

Show Kernel Estimate

Lookback Window

Relative Weighting

Regression Level

Enhance Kernel Smoothing

Lag

Kernel Settings Help

Select a kernel setting to view its help information

Core Concepts

Why Use Kernel Settings?

• Adaptive Smoothing: More nuanced than an SMA/EMA, capturing subtle directional shifts.
• Noise Reduction: Helps filter out random price spikes or dips.

Quick Reference Table

Setting	Default	Purpose	Range (Recommended)
Trade with Kernel	True	Activates kernel-based trend	On/Off
Show Kernel Estimate	True	Displays the kernel line	On/Off
Enhance Kernel Smooth	False	Reduces color jitter via crossovers	On/Off
Lookback Window	8	Bars used for kernel calc	3–50
Relative Weighting	8.0	Emphasis on recent vs. older data	0.25–25
Regression Level	25	Start bar for regression	2–25
Lag	2	Delay in crossover detection	1–2

Configuration Parameters

Core Settings

1. Trade with Kernel

   • Toggles kernel-based logic for signals instead of raw price or standard MAs.
   • Often more stable in choppy markets.

2. Show Kernel Estimate

   • Visual feedback on the chart for real-time trend interpretation.

3. Enhance Kernel Smoothing

   • Additional logic (crossovers/thresholds) to reduce frequent color flips.

Advanced Parameters

4. Lookback Window

   • Shorter = more reactive, risk of overfitting. Longer = smoother, but can lag.

5. Relative Weighting

   • High weighting emphasizes recent bars, capturing fresh market dynamics.
   • Low weighting spreads the influence over a broader time window.

6. Regression Level

   • Skips early bars that might distort baseline, focusing on more stable data from bar X onward.

7. Lag

   • Deliberate delay to confirm crossovers and reduce whipsaws.

Trading Configurations

Reactive Intraday

• Lookback: ~5
• Relative Weighting: 5–8
• Regression Level: 2–5
• Lag: 1

Steady Swing

• Lookback: 10–12
• Relative Weighting: ~8 (default)
• Regression Level: 20–25
• Lag: 2

Optimization Tips

Best Practices

• Layer with Filters: Combine kernel-based direction with ADX/Regime filters.
• Conditional Adjustments: In high volatility, shorten the window. In calmer periods, lengthen.
• Iterate Regularly: Document changes & observe improvements.

Key Takeaway: Fine-tuning kernel settings gives you powerful control over how quickly the model senses and reacts to directional changes, balancing responsiveness with noise reduction.

Example Configurations

Smooth Trend Following:

• Trade with Kernel: True - Reasoning: Enables kernel-based trend estimation, providing a more adaptive approach than simple moving averages.
• Show Kernel Estimate: True - Reasoning: Provides visual confirmation of the trend, allowing for quick assessment of market direction.
• Enhance Kernel Smoothing: True - Reasoning: Reduces noise and provides a clearer trend signal, minimizing whipsaws and false signals.
• Lookback Window: 12 - Reasoning: Captures a broader range of price action for smoother trend estimation, reducing sensitivity to short-term fluctuations.
• Relative Weighting: 5.0 - Reasoning: Balances responsiveness and stability in trend detection, giving more weight to recent data while still considering historical context.
• Regression Level: 20 - Reasoning: Skips initial volatile bars for a more stable baseline, focusing on more established trends.
• Lag: 2 - Reasoning: Reduces whipsaws and confirms crossover signals, improving the reliability of trend entries. This configuration is designed for Backtest Optimizers and Algorithm Curious Learners interested in smooth trend estimation. It employs enhanced kernel smoothing and a longer lookback window to create a smoother, more stable kernel estimate, ideal for trend-following strategies and reducing noise. A higher Regression Level skips initial bars for a better fit, and the Lag setting helps to confirm crossover signals.

Responsive Intraday Trading:

• Trade with Kernel: True - Reasoning: Enables kernel-based trend estimation, providing a more adaptive approach than simple moving averages.
• Show Kernel Estimate: True - Reasoning: Provides visual confirmation of the trend, allowing for quick assessment of market direction.
• Enhance Kernel Smoothing: False - Reasoning: Allows for quicker reactions to price changes, essential for scalping.
• Lookback Window: 5 - Reasoning: Captures short-term price movements for scalping, increasing responsiveness to intraday fluctuations.
• Relative Weighting: 10.0 - Reasoning: Emphasizes recent price action for responsiveness, prioritizing the most recent data.
• Regression Level: 5 - Reasoning: Fits the kernel tighter to recent data, capturing short-term trends.
• Lag: 1 - Reasoning: Minimizes delay in crossover detection for timely entries, crucial for scalping. This configuration is designed for Performance Tweakers and Reactive Intraday Traders focused on responsiveness. It uses a shorter lookback window and disables kernel smoothing for a more reactive kernel estimate, suitable for intraday trading and capturing quick price movements. A lower Regression Level fits the kernel tighter to recent data, and a lower Lag minimizes delay in crossover detection.

Long-Term Trend Analysis:

• Trade with Kernel: True - Reasoning: Enables kernel-based trend estimation, providing a more adaptive approach than simple moving averages.
• Show Kernel Estimate: True - Reasoning: Provides visual confirmation of the trend, allowing for quick assessment of market direction.
• Enhance Kernel Smoothing: True - Reasoning: Reduces noise and provides a clearer trend signal, minimizing whipsaws and false signals.
• Lookback Window: 20 - Reasoning: Captures a broader range of price action for smoother trend estimation, reducing sensitivity to short-term fluctuations.
• Relative Weighting: 2.0 - Reasoning: Emphasizes longer-term trends and reduces sensitivity to short-term fluctuations, providing a more stable view of the market.
• Regression Level: 30 - Reasoning: Further stabilizes the kernel by focusing on later, more stable data, reducing the impact of initial volatility.
• Lag: 2 - Reasoning: Reduces whipsaws and confirms crossover signals, improving the reliability of trend entries. This configuration is designed for Fixed Income/Conservative Traders and Algorithm Curious Learners interested in long-term trend analysis. It employs a longer lookback window and lower relative weighting to emphasize longer-term trends and reduce sensitivity to short-term fluctuations. A higher Regression Level further stabilizes the kernel by focusing on later, more stable data.