Leveraging Data Science for Robust Trading Strategies

Our third guest in Market Masters is Martí Castany, a Spanish Quant Researcher and Portfolio Manager. With an Electrical and Hardware Engineering background and more than a decade of experience in the systematic investment sector, Martí has focused on developing statistical models, data analytics and cleansing, and quantitative risk management.

Beginnings in Algorithmic Trading

Martí's knowledge of data analytics is entirely self-taught, perfected through his work with various companies. He started with a startup analysing massive data sets from brokerage firms to predict client profitability. Later, at a boutique consultancy, he also handled extensive data processing. Since then, his focus has always been on translating theoretical concepts into practical, real-world applications.

In 2014, Martí began trading futures discretionarily and later ventured into cryptocurrencies. Over time, he realised that his trading decisions were based solely on historical market behaviour rather than actual data. That’s when he decided to start incorporating a scientific approach into his decision-making, being systematic, and then coding the logic into algorithms that could be executed.

Building a Robust Trading Framework

Today, Martí emphasises the importance of having a robust framework for developing trading strategies. In our recent conversation, he explained that having a process to follow is fundamental: “It gives you enough confidence to maintain your strategy if/when it has a bad time.”

Martí also explained the process he follows before designing a trading strategy, starting with how to generate a new trading idea, hypothesis testing, use of synthetic data, and everything necessary before confidently going live.

Having a Solid Infrastructure

Another cornerstone of Martí’s approach is the importance he places on having an infrastructure that supports your strategy: “You can have the best strategy in the world, but if you have a bad infrastructure, it won't serve you.” He points out that many algorithmic traders tend to overlook the significance of infrastructure. However, the underlying elements, such as the code, hardware, and server, are as crucial as the strategy itself.

Recognising this, Martí and his team, frustrated by human errors when transitioning research code into production, spent six months building their own trading infrastructure. This event-driven, backtesting, and live trading framework now resides in a GIT repository, allowing different team members to work on various parts of the framework simultaneously.

Managing Trading Uncertainty

Lastly, Martí explained the relevance of understanding and managing uncertainty in trading. He uses non-parametric bootstrapping to measure it, allowing him to simulate thousands of scenarios and assess the robustness of his strategies. This method helps in making informed decisions about asset selection and portfolio composition, reducing trading risks.

For those new to data science in trading, Martí’s top advice is:

In conclusion, Martí Castany’s journey and insights offer a comprehensive guide for anyone looking to leverage data science in trading. His approach emphasises integrating scientific rigour into a well-thought process to build a robust trading strategy.

Don’t miss this episode if you are interested in learning how to integrate data science to further empower your trading and stay tuned for more episodes of Market Masters, where we continue to explore innovative strategies and insights from seasoned traders around the world.