Savitzky-Golay Hampel Filter | AlphaNattSavitzky-Golay Hampel Filter | AlphaNatt
A revolutionary indicator combining NASA's satellite data processing algorithms with robust statistical outlier detection to create the most scientifically advanced trend filter available on TradingView.
"This is the same mathematics that processes signals from the Hubble Space Telescope and analyzes data from the Large Hadron Collider - now applied to financial markets."
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
🚀 SCIENTIFIC PEDIGREE
Savitzky-Golay Filter Applications:
NASA: Satellite telemetry and space probe data processing
CERN: Particle physics data analysis at the LHC
Pharmaceutical: Chromatography and spectroscopy analysis
Astronomy: Processing signals from radio telescopes
Medical: ECG and EEG signal processing
Hampel Filter Usage:
Aerospace: Cleaning sensor data from aircraft and spacecraft
Manufacturing: Quality control in precision engineering
Seismology: Earthquake detection and analysis
Robotics: Sensor fusion and noise reduction
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
🧬 THE MATHEMATICS
1. Savitzky-Golay Filter
The SG filter performs local polynomial regression on data points:
Fits a polynomial of degree n to a sliding window of data
Evaluates the polynomial at the center point
Preserves higher moments (peaks, valleys) unlike moving averages
Maintains derivative information for true momentum analysis
Originally published in Analytical Chemistry (1964)
Mathematical Properties:
Optimal smoothing in the least-squares sense
Preserves statistical moments up to polynomial order
Exact derivative calculation without additional lag
Superior frequency response vs traditional filters
2. Hampel Filter
A robust outlier detector based on Median Absolute Deviation (MAD):
Identifies outliers using robust statistics
Replaces spurious values with polynomial-fitted estimates
Resistant to up to 50% contaminated data
MAD is 1.4826 times more robust than standard deviation
Outlier Detection Formula:
|x - median| > k × 1.4826 × MAD
Where k is the threshold parameter (typically 3 for 99.7% confidence)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
💎 WHY THIS IS SUPERIOR
vs Moving Averages:
Preserves peaks and valleys (critical for catching tops/bottoms)
No lag penalty for smoothness
Maintains derivative information
Polynomial fitting > simple averaging
vs Other Filters:
Outlier immunity (Hampel component)
Scientifically optimal smoothing
Preserves higher-order features
Used in billion-dollar research projects
Unique Advantages:
Feature Preservation: Maintains market structure while smoothing
Spike Immunity: Ignores false breakouts and stop hunts
Derivative Accuracy: True momentum without additional indicators
Scientific Validation: 60+ years of academic research
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
⚙️ PARAMETER OPTIMIZATION
1. Polynomial Order (2-5)
2 (Quadratic): Maximum smoothing, gentle curves
3 (Cubic): Balanced smoothing and responsiveness (recommended)
4-5 (Higher): More responsive, preserves more features
2. Window Size (7-51)
Must be odd number
Larger = smoother but more lag
Formula: 2×(desired smoothing period) + 1
Default 21 = analyzes 10 bars each side
3. Hampel Threshold (1.0-5.0)
1.0: Aggressive outlier removal (68% confidence)
2.0: Moderate outlier removal (95% confidence)
3.0: Conservative outlier removal (99.7% confidence) (default)
4.0+: Only extreme outliers removed
4. Final Smoothing (1-7)
Additional WMA smoothing after filtering
1 = No additional smoothing
3-5 = Recommended for most timeframes
7 = Ultra-smooth for position trading
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
📊 TRADING STRATEGIES
Signal Recognition:
Cyan Line: Bullish trend with positive derivative
Pink Line: Bearish trend with negative derivative
Color Change: Trend reversal with polynomial confirmation
1. Trend Following Strategy
Enter when price crosses above cyan filter
Exit when filter turns pink
Use filter as dynamic stop loss
Best in trending markets
2. Mean Reversion Strategy
Enter long when price touches filter from below in uptrend
Enter short when price touches filter from above in downtrend
Exit at opposite band or filter color change
Excellent for range-bound markets
3. Derivative Strategy (Advanced)
The SG filter preserves derivative information
Acceleration = second derivative > 0
Enter on positive first derivative + positive acceleration
Exit on negative second derivative (momentum slowing)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
📈 PERFORMANCE CHARACTERISTICS
Strengths:
Outlier Immunity: Ignores stop hunts and flash crashes
Feature Preservation: Catches tops/bottoms better than MAs
Smooth Output: Reduces whipsaws significantly
Scientific Basis: Not curve-fitted or optimized to markets
Considerations:
Slight lag in extreme volatility (all filters have this)
Requires odd window sizes (mathematical requirement)
More complex than simple moving averages
Best with liquid instruments
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
🔬 SCIENTIFIC BACKGROUND
Savitzky-Golay Publication:
"Smoothing and Differentiation of Data by Simplified Least Squares Procedures"
- Abraham Savitzky & Marcel Golay
- Analytical Chemistry, Vol. 36, No. 8, 1964
Hampel Filter Origin:
"Robust Statistics: The Approach Based on Influence Functions"
- Frank Hampel et al., 1986
- Princeton University Press
These techniques have been validated in thousands of scientific papers and are standard tools in:
NASA's Jet Propulsion Laboratory
European Space Agency
CERN (Large Hadron Collider)
MIT Lincoln Laboratory
Max Planck Institutes
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
💡 ADVANCED TIPS
News Trading: Lower Hampel threshold before major events to catch spikes
Scalping: Use Order=2 for maximum smoothness, Window=11 for responsiveness
Position Trading: Increase Window to 31+ for long-term trends
Combine with Volume: Strong trends need volume confirmation
Multiple Timeframes: Use daily for trend, hourly for entry
Watch the Derivative: Filter color changes when first derivative changes sign
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
⚠️ IMPORTANT NOTICES
Not financial advice - educational purposes only
Past performance does not guarantee future results
Always use proper risk management
Test settings on your specific instrument and timeframe
No indicator is perfect - part of complete trading system
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
🏆 CONCLUSION
The Savitzky-Golay Hampel Filter represents the pinnacle of scientific signal processing applied to financial markets. By combining polynomial regression with robust outlier detection, traders gain access to the same mathematical tools that:
Guide spacecraft to other planets
Detect gravitational waves from black holes
Analyze particle collisions at near light-speed
Process signals from deep space
This isn't just another indicator - it's rocket science for trading .
"When NASA needs to separate signal from noise in billion-dollar missions, they use these exact algorithms. Now you can too."
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Developed by AlphaNatt
Version: 1.0
Release: 2025
Pine Script: v6
"Where Space Technology Meets Market Analysis"
Not financial advice. Always DYOR
Penunjuk Pine Script®






















