Jo­nas Stri­aukas

In my research, I develop statistical and machine learning methods for high-dimensional, heavy-tailed, and dependent data, with applications in economics and finance. My work focuses on transforming complex, noisy datasets into reliable, real-time insights for decision-making. More recently, my research has focused on two areas: developing testing procedures for tail (i.e., extreme) outcomes, and developing sequential global testing frameworks that provide rigorous statistical guarantees.

In macroeconomic applications, I build nowcasting tools that use mixed-frequency (MIDAS) and machine learning techniques to track economic activity in real time. These methods are robust to publication lags, data revisions, and structural breaks—helping central banks and firms respond more quickly to sudden shifts in the economy.

In finance, I design tests for risk premia and implementable portfolio strategies using high-frequency market and alternative data (such as news articles).

My approaches account for realistic trading constraints, high dimensionality, and multiple testing challenges. I promote open and reproducible research and teach evidence-based analytics, statistics, and machine learning. I also collaborate with Euro area institutions to improve the speed, fairness, and transparency of decisions under uncertainty.