"Running an automated systematic hedge fund is not rocket science. But there are so many things that can go wrong. Solid engineering and interdisciplinary knowledge are the foundation of a reliable system.”


Part 2

of our series about our interdisciplinary specialists and their contributions at Tom Capital

Dr. Christian Gloor leads the quantitative machine learning research at Tom Capital. With more than 15 years of relevant experience in the finance industry and an education in computer science, he is able to glue together the different components of Tom Capital.

Before joining Tom Capital in March 2017 Christian was Senior Quantitative Developer at Man Group’s systematic division AHL. His team was responsible for developing the next generation of strategies, such as a smart risk parity product that became AHL’s best performing fund. Earlier, as CTO of the company that built and operated the second-largest stock exchange of Switzerland, BX Swiss, he was responsible for the overall design of the electronic trading platform.

Christian holds a PhD from ETH Zürich in distributed machine learning. His thesis combined self-learning agent behavior with a high-speed, large-scale simulation system. Key aspects were decision making on uncertain, incomplete and temporal knowledge, and experiment design and validation.

His goal is to manage funds using sophisticated quantitative models built on a state-of-the-art research and trading infrastructure. Being familiar with all aspects of a systematic hedge fund, Christian was able to steer the development of Tom Capital's processes and infrastructure to an institutional level. He contributed best practices like out-of-sample and walk-forward backtesting, cross-validation, volatility-based capital allocation, and execution algorithms. Built on the original model, the fund returns became smoother, risk-controlled, and more predictable.

The existing, manually designed global macro models are now augmented with various modern machine learning techniques, such as genetic algorithms, random forests, K-nearest neighbors, and deep networks. To support this process, a research framework with access to a new in-house compute cluster has been built. Christian developed a modular Python SDK (Software Development Kit) containing the basic building blocks for quantitative research and trading. This paved the way for a stable, fully automated trading system.

July 30  2019