Machine Learning Researcher - High Frequency Trading

Role Overview

Join a leading global high-frequency trading firm expanding its systematic equities platform in Singapore. This position focuses on developing machine learning techniques for rapid, short-term price predictions within Asian equity markets characterized by fast signal decay, limited capacity, and critical execution quality.

Key Responsibilities

• Create machine learning models aimed at forecasting price movements over various intervals including tick-level, sub-second, and intraday horizons.
• Analyze order book behavior such as queue position, order flow imbalances, liquidity provisioning, and the risks of adverse selection, transforming these findings into actionable trading signals.
• Develop features extracted directly from comprehensive order book data, including trade-by-trade and nanosecond-precision timestamps.
• Design models mindful of real-world operational limits like latency constraints, exchange throttling, fill probabilities, market impact, and transaction costs.
• Manage the entire research lifecycle: from initial hypothesis formulation through production deployment and continual live monitoring of model performance degradation.
• Collaborate closely with quantitative developers and execution engineers, ensuring seamless integration of research and infrastructure.

Candidate Profile

• Possession of a PhD or Master's degree in Machine Learning, Statistics, Computer Science, Mathematics, Physics, or a related quantitative field from a reputable academic institution.
• Demonstrated expertise in applied machine learning, particularly with low signal-to-noise, high-frequency financial data, including familiarity with online learning, regularization methods, and adaptation to regime changes.
• Strong proficiency in Python and advanced skills in C++, given the latency-critical nature where research programming closely aligns with production code.
• Experience handling high-frequency market datasets at scale, including Level 2/Level 3 order book reconstruction, tick data processing, and disciplined timestamp management.
• A methodical approach to backtesting high-frequency trading tactics, incorporating realistic simulations for fills and slippage effects.