quant-research-system

quant-research-system is a research-oriented quantitative trading project focused on building and evaluating an equity long-only strategy pipeline.

The repository is designed to demonstrate end-to-end quant engineering capabilities, including:

• data preparation and universe construction
• feature engineering workflows
• alpha model training and evaluation
• portfolio backtesting and risk analysis
• strategy parameter sweeps and reporting

Project Goal

The goal of this project is to provide a modular framework for turning market data into tradable portfolio decisions, then validating performance through reproducible backtests and diagnostics.

What This Repository Highlights

• System design: clear separation between data, features, modeling, and backtesting modules
• Research workflow: iterative experiments for strategy tuning and risk/return trade-offs
• Engineering practices: script-based pipelines, structured outputs, and test coverage

Repository Structure

• core/data_build: data ingestion and canonicalization
• core/universe: tradable universe construction rules
• core/features: feature pipeline components
• core/models/Xgboost: alpha modeling, prediction, and evaluation
• core/backtest: strategy simulation, parameter sweeps, and reports
• dq: data quality checks
• tests: validation and guardrail tests

Notes on Public Version

This public repository is intentionally sanitized for portfolio and interview purposes. Certain strategy-sensitive details (for example, proprietary alpha definitions and some production configurations) are omitted or abstracted.

The focus of this version is to showcase the architecture, workflow, and implementation quality of a practical quant research system.