Kent Directional Filter🧭 Kent Directional Filter
Author: GabrielAmadeusLau
Type: Filter
📖 What It Is
The Kent Directional Filter is a directionality-sensitive smoothing tool inspired by the Kent distribution, a probability model used to describe directional and elliptical shapes on a sphere. In this context, it's repurposed for analyzing the angular trajectory of price movements and smoothing them for actionable insights.
It’s ideal for:
Detecting directional bias with probabilistic weighting
Enhancing momentum or trend-following systems
Filtering non-linear price action
🔬 How It Works
Price Angle Estimation:
Computes a rough angular shift in price using atan(src - src ) to estimate direction.
Kent Distribution Weighting:
κ (kappa) controls concentration strength (how sharply it prefers a direction).
β (beta) controls ellipticity (bias toward curved vs. linear moves).
These parameters influence how strongly the indicator favors movements at ~45° angles, simulating a directional “lens.”
Smoothing:
A Simple Moving Average (SMA) is applied over the raw directional probabilities to reduce noise and highlight the underlying trend signal.
⚙️ Inputs
Source: Price series used for angle calculation (default: close)
Smoothing Length: Window size for the moving average
Pi Divisor: Pi / 4 would be 45 degrees, you can change the 4 to 3, 2, etc.
Kappa (κ): Controls how focused the directionality is (higher = sharper filter)
Beta (β): Adds curvature sensitivity; higher values accentuate asymmetrical moves
🧠 Tips for Best Results
Use κ = 1–2 for moderate directional filtering, and β = 0.3–0.7 for smooth elliptical bias.
Combine with volume-based indicators to confirm breakout strength.
Works best in higher timeframes (1h–1D) to capture macro directional structure.
I might revisit this.
