EventTrendLab

Research Framework for Signal Selection, Portfolio Construction, and Quantitative Strategy Evaluation

---

Overview

EventTrendLab is a research framework designed to evaluate how different stock-selection layers affect the performance of the same underlying trading strategy.

The framework separates:

• Signal Generation
• Selection Models
• Portfolio Construction
• Risk Management
• Benchmark Analysis

This allows researchers to isolate the contribution of the selection layer while keeping all other components unchanged.

---

Research Question

Can portfolio performance be improved through better trade selection while keeping the underlying strategy unchanged?

This project evaluates the impact of different selection approaches on:

• CAGR
• Sharpe Ratio
• Maximum Drawdown
• Capital Efficiency
• Exposure Utilization

---

Research Results

No Selection (Baseline)

Characteristics:

• Highest exposure
• Highest drawdown
• Weakest risk-adjusted return
• No capital allocation optimization

---

Rule-Based Selection

Characteristics:

• Improved signal quality
• Lower drawdown
• Better Sharpe ratio
• More efficient capital deployment

---

Machine Learning Selection

Characteristics:

• Strongest long-term performance
• Improved portfolio concentration
• Better capital efficiency
• Superior risk-adjusted return

---

Framework Architecture

TradingView Strategy Exports
                │
                ▼
        Signal Loader
                │
                ▼
        Selection Layer
                │
        ┌───────┴────────┐
        │                │
        ▼                ▼
 Rule-Based      Machine Learning
 Selection           Selection
        │                │
        └───────┬────────┘
                │
                ▼
         Position Sizing
                │
                ▼
      Portfolio Aggregation
                │
                ▼
       Mark-to-Market Engine
                │
                ▼
        Performance Analysis
                │
                ▼
      Benchmark Comparison

---

Key Components

Market Panel Engine

Builds a cross-sectional market panel from historical stock data.

Example fields:

• Open
• High
• Low
• Close
• Volume
• Turnover
• Sector Information

---

Signal Ingestion

Converts TradingView strategy exports into standardized portfolio events.

Output:

• Entry Events
• Exit Events
• Position Records

---

Selection Layer

Supports multiple selection methodologies:

• No Selection
• Rule-Based Selection
• Machine Learning Selection

Selection models are evaluated independently while maintaining the same underlying trading strategy.

---

Portfolio Engine

Portfolio-level simulation including:

• Position Sizing
• Exposure Control
• Capital Allocation
• Daily Mark-to-Market Accounting

---

Benchmark Analysis

Performance comparison against:

• Buy & Hold Benchmark
• Exposure-Matched Benchmark

Metrics:

• CAGR
• Sharpe Ratio
• Max Drawdown
• Exposure Statistics

---

Repository Structure

EventTrendLab/

├── backtest/
│   ├── analyze_portfolio.py
│   └── position_sizing.py
│
├── ingestion/
│   └── tv_loader.py
│
├── selection/
│   ├── config.py
│   └── demo_selection.py
│
├── data/
│   └── _sample/
│       ├── market_panel_sample.csv
│       └── tv_exports/
│
├── images/
│   ├── none.jpg
│   ├── rule.jpg
│   └── ai.jpg
│
├── exports/
│
├── main.py
│
└── README.md

---

Quick Start

Clone Repository

git clone https://github.com/YOUR_USERNAME/EventTrendLab-public.git
cd EventTrendLab-public

Install Dependencies

pip install pandas numpy matplotlib

Run Demo

python main.py

Outputs:

exports/

portfolio_equity_curve.csv
portfolio_summary.csv
portfolio_vs_0050.png
portfolio_exposure_stats.png

---

Public Version

This repository contains a simplified public implementation of the research framework.

The production research environment includes additional components that are intentionally excluded:

• Proprietary alpha generation modules
• Production datasets
• Advanced feature engineering
• Extended stock universes
• Research infrastructure
• Portfolio optimization modules

The public version is intended to demonstrate framework architecture and research workflow only.

---

Tech Stack

• Python
• Pandas
• NumPy
• Matplotlib

Research Topics:

• Quantitative Research
• Portfolio Construction
• Risk Management
• Relative Strength Selection
• Machine Learning Selection
• Systematic Trading

---

Disclaimer

This repository is provided for research and educational purposes only.

Nothing in this repository should be interpreted as investment advice or a recommendation to buy or sell any financial instrument.

Past performance does not guarantee future results.

---

Author

Independent Quant Researcher

Research Interests:

• Portfolio Construction
• Quantitative Investing
• Systematic Trading
• Selection Models
• Risk Management