Market Microstructure Study

A compact quant research project on how volatility and liquidity affect trading frictions.

Built with synthetic data to demonstrate research structure, reproducibility, and clear interpretation.

quant research market microstructure synthetic data notebook workflow reproducible analysis

Preview

Overview

A compact research workflow built around one question: how volatility and liquidity shape trading frictions in a stylized synthetic market.

Research Question

How do volatility and liquidity affect a simple spread proxy?

Hypothesis

• Higher volatility should be associated with wider spreads.
• Lower liquidity should be associated with wider spreads.
• The widening effect should be strongest during stress regimes.

Quick Start

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
python src/generate_data.py
python src/analysis.py

Methods

The study generates cross-sectional daily data for synthetic instruments across different liquidity tiers and market-cap buckets. It then:

1. simulates returns, volatility, turnover, and spread proxies
2. labels observations into calm versus stress regimes
3. estimates a simple OLS relationship between spread, volatility, and log turnover
4. compares spread behavior across liquidity tiers and stress regimes
5. saves charts and summary outputs for portfolio presentation

Key Outputs

After running the scripts, you should see:

• data/synthetic_microstructure_data.csv
• results/summary_metrics.csv
• results/regression_summary.csv
• results/key_findings.md
• figures/spread_vs_volatility.png
• figures/spread_by_regime_and_liquidity.png
• figures/spread_distribution_by_liquidity.png

Example Findings From The Synthetic Sample

• Average spread proxy is wider in stress regimes than in calm regimes: 31.35 bps versus 24.00 bps
• Low-liquidity names show materially wider spreads than high-liquidity names: 38.07 bps versus 18.92 bps
• In the simple OLS, realized volatility loads positively on spread: +283.17
• Log turnover loads negatively on spread: -8.71
• The stress indicator remains positive even after controlling for volatility and turnover: +4.55

Project Structure

• data/: generated synthetic dataset
• figures/: charts created by the analysis script
• notebooks/: notebook-style walkthrough starter
• results/: summary tables and key findings
• src/: data generation and analysis code

Why It Fits The Portfolio

This project is designed to complement a dashboard portfolio piece. Instead of showcasing interface design, it showcases research thinking:

• forming a clear hypothesis
• building a reproducible dataset
• running a simple empirical analysis
• summarizing findings in a compact and readable way

Limitations

• The dataset is synthetic and intended for demonstration rather than live inference.
• The spread proxy is stylized, not a direct market microstructure measurement.
• The regression is deliberately simple and meant to be easy to explain in an interview.