Most systematic strategies fail because they rely on time-averaging indicators (Moving Averages, RSI, MACD). In a non-stationary system like the limit order book, averaging past data creates a terminal phase lag. By the time your indicator signals a "trend," the institutional move is already exhausted.
This repository demonstrates a Kinematic approach to price action. Instead of treating price as a series of static candles, we treat it as a continuous physical object in motion.
-
Savitzky-Golay Filtering: Replacing lagging MAs with polynomial-fit derivatives to extract instantaneous Velocity (
$v$ ) and Acceleration ($a$ ). - Causal Hilbert Transforms: Utilizing reflective padding to extract the instantaneous phase of the market cycle without "end-point repainting."
-
Shannon Entropy (
$H$ ): Measuring the Information Density of the signal to identify High Chaos environments where trend-following is statistically suicidal. - Hidden Markov Models (HMM): A Bayesian framework that synthesizes Kinematics and Entropy to decode the hidden structural regime of the market (e.g., Bull Drift vs. Kinematic Dip).
The logic in this repo is a subset of the Impulse-X Engine, a high-speed terminal environment optimized for disciplined traders who require objective math over gut feelings.
- Strictly Causal: No look-ahead bias. No repainting.
- Signal Demodulation: Separating structural alpha from high-entropy noise.
- Regime Awareness: Automatically adjusting bias based on the HMM transition matrix.
For the full mathematical derivation, LaTeX frameworks, and terminal implementation specs, visit the official hub:
👉 Impulse-X Documentation & Technical Manual
📊 TradingView User? I've included a free Pine Script version of the Kinematic Velocity engine. Copy the code from the /pinescript folder and paste it into your editor. Secure your Alpha Seat on Gumroad