It is challenging to understand how to model external shocks when trading financial markets. However, in recent years, it has become particularly notable that these risks, such as Brexit, the election of Trump, or coronavirus can greatly impact markets. Hence, we need to have a way to model them.
In this paper, Cuemacro investigates the Thorfinn Sensitivity Index (TSI) which quantifies event risks related to geopolitics and related areas. The aim is to provide a quantifiable output for these risks. Such an output can be easier to interpret for market participants, whether they are discretionary or systematic traders. The TSI uses machine learning and natural language processing to parse through textbased sources which are in the political and geo-economic arena, in particular, it parses through think thanks, research centres, and certain social media activity. Experts are also involved at various stages in the process to score the various inputs once they have been aggregated into a more digestible form.
Research shows that, historically when the index flags increases in risk this tends to be accompanied by an underperformance of risky assets and outperformance of safe haven assets. Cuemacro uses the TSI index to create systematic trading strategies for macro-based assets. Its macro trading basket strategy which uses signals based on TSI has annualized returns of 14.8% and risk adjusted returns of 1.32 over the past 2 years, outperforming a passive strategy.