Fisher MPz

11월 8일

Fisher MPz - Multi-Period Z-Score Fisher Transform

Overview

An enhanced Fisher Transform that uses multi-period analysis and improved statistical methods to provide more reliable trading signals with the goal of fewer false positives.

Evolution Beyond Traditional Fisher Transform

While the classic Fisher Transform uses simple price normalization and basic smoothing, Fisher MPz introduces several key enhancements:
- Multi-period composite instead of single timeframe analysis
- Robust z-score normalization using median/MAD rather than mean/standard deviation
- Winsorization to handle outliers and price spikes
- Dynamic clipping that adapts to market volatility
- Kalman filtering for superior noise reduction vs. traditional EMA smoothing

These improvements result in cleaner signals, better adaptability to different market conditions, handles trending markets without over-saturation at extreme values, and reduced false signals compared to the standard Fisher Transform.

Key Features

Multi-Period Analysis
- Three Timeframe Approach: Simultaneously analyzes short (default 8), medium (default 13), and long (default 26) periods
- Weighted Composite: Combines all three periods using customizable weights for optimal signal generation
- Individual Period Display: Optional visualization of each period's Fisher Transform for deeper analysis

Advanced Statistical Methods

Robust Z-Score Calculation
- Uses median and MAD (Median Absolute Deviation) instead of mean and standard deviation
- More resistant to outliers and extreme price movements
- Provides stable normalization across varying market conditions

Winsorization
- Caps extreme price values at specified percentiles (default 5th and 95th)
- Reduces the impact of price spikes and anomalies
- Configurable lookback period for threshold calculation

Dynamic Z-Score Clipping
- Automatically adjusts clipping levels based on recent volatility
- Tighter bounds in calm markets (0.05) for precision
- Wider bounds in volatile markets (0.2) to capture significant moves
- Uses ATR-based volatility measurement

Kalman Filter Smoothing
- Optional advanced noise reduction using Kalman filtering
- Superior to traditional EMA smoothing for optimal signal extraction
- Configurable process noise (Q) and measurement noise (R) parameters
- Fallback to traditional smoothing factor available

How to Use

Basic Interpretation
- Above Zero: Bullish momentum
- Below Zero: Bearish momentum
- Extreme Values: Potential overbought/oversold conditions
- Crossovers: Entry/exit signals when composite crosses trigger line

Customizable Settings

Periods: Adjust based on your trading timeframe
- Lower values (3-10): More sensitive, suitable for scalping
- Medium values (10-20): Balanced for swing trading
- Higher values (20-50): Smoother for position trading

Weights: Customize responsiveness
- Increase short weight: More reactive to recent price changes
- Increase long weight: More stability and trend confirmation

Kalman Settings
- Lower Q (0.001-0.02): Smoother, more filtered signals
- Higher Q (0.02-0.1): More responsive to price changes
- Lower R (0.01-0.05): Trust data more, less filtering
- Higher R (0.1-1.0): More skeptical of data, more smoothing

5일 전

릴리즈 노트

Added new visual style