Market Microstructure Simulator

Problem

Modern market dynamics are driven by hidden interactions between market makers, informed traders, institutional flow, retail agents and execution algorithms. From observable trades, cancellations and quote updates alone, the underlying composition of the market is only partially visible.

Approach

The simulator generates large synthetic markets with controlled agent populations, then uses the resulting order flow, placement dynamics, cancellations and liquidity behavior as training data for inverse models that infer latent agent distributions.

Technical Focus

• Event-level limit order book simulation.
• Realistic matching, market impact and order flow dynamics.
• Stochastic volatility and liquidity regimes.
• Structured execution logs for machine learning analysis.

Agents

• Market makers and liquidity providers.
• Informed traders and directional flow.
• Institutional execution algorithms.
• Retail and noise-driven participants.

Research Use

The framework serves both as a controlled environment for market microstructure experiments and as a synthetic data generation engine for AI-based inverse market modelling.

Current Objectives

• Inverse inference of latent agent distributions.
• Reinforcement learning and AI-driven market analysis.
• Fragmented liquidity and multi-venue interactions.
• Calibration to empirical stylized facts.

Long-Term Goal

Build a research-grade framework linking observable microstructure dynamics to the hidden behavioral composition of financial markets.

Stack