title	Market DNA Engine
emoji	📊
colorFrom	blue
colorTo	indigo
sdk	streamlit
app_file	0_Home.py
pinned	false

Market DNA Engine

Institutional Digital Signal Processing for Non-Lagging Alpha Extraction

The Market DNA Engine is a professional-grade strategy analysis platform that utilizes Digital Signal Processing (DSP) to decompose market price action into orthogonal frequency components. This architecture separates structural macro trends from stochastic algorithmic noise, providing a non-lagging foundation for institutional decision-making.

---

Live Demo

1. Problem Identification

Traditional technical indicators (Moving Averages, RSI, MACD) are fundamentally reactive. They rely on lookback windows that introduce phase lag-the delay between a market event and the indicator's signal. In modern high-frequency regimes, this delay leads to "whipsaw" entries and delayed exits, eroding alpha. Furthermore, raw price data is "noisy," combining long-term value shifts with short-term liquidity distortions.

2. Innovative Approach

Market DNA solves the lag problem by shifting analysis from the Time Domain to the Frequency Domain. By using Orthogonal Wavelet Decomposition (MRA) and Synchrosqueezing, the engine extracts specific "Market Rhythms" (cycles) without the temporal distortion inherent in traditional filters. This allows traders to identify structural support and cyclical exhaustion before they manifest in lagging indicators.

3. Technical Implementation

• Signal Extraction: Multiresolution Analysis (MRA) using Daubechies (db4) and Symlets (sym8) wavelets.
• Resonance Tracking: Cross-Wavelet Coherence (CWT) for tracking leading/lagging relationships between assets (e.g., BTC as a leading indicator for SPY).
• Causal Inference: Vector Autoregression (VAR) based Spectral Granger Causality to identify directional information flow.
• Risk Management: Integrated Backtesting suite with Kelly Criterion capital allocation and OOS (Out-of-Sample) validation.

4. Usability and User Experience

The platform features a high-fidelity Slate-Carbon Tactical HUD, designed for high-density information display. It provides clinical "Execution Playbooks" that translate complex DSP waveforms into actionable 1-2-3 trading instructions.

5. Scalability and Feasibility

The engine is built on a modular data-agnostic layer. It defaults to public Yahoo Finance data for accessibility but is fully bridged for institutional APIs (Alpaca, Polygon, Interactive Brokers). The DSP logic is optimized via NumPy/SciPy for sub-second analysis on standard commodity hardware.

---

Technologies Used

• Language: Python 3.10+
• Frontend: Streamlit (Framework), Vanilla CSS (Custom HUD)
• DSP Core: PyWavelets, ssqueezepy, pycwt
• Analysis: Statsmodels, Pandas, NumPy
• Visualization: Plotly (Interactive Spectral Charts)
• Deployment: Hugging Face Spaces, GitHub Actions

---

Setup Instructions

1. Clone the Repository:

   git clone https://github.com/JoshChomps/Finance_using_DSP.git
   cd Finance_using_DSP

2. Install Dependencies:

   pip install -r requirements.txt

3. Run Locally:

   streamlit run 0_Home.py

---

Team Details

• Josh Chomiak: Lead Engineer & Quantitative Architect
  • Contribution: End-to-end development of the DSP engine, HUD interface, and backtesting framework.

---

Institutional Standard | Algofest 2026 | Systematic Alpha.