Board Info Session
📍Location: TS53 C-1.03
🗓️Date: Tuesday, 17th March, 2026
🕧Time: 18:30
Interested in joining the board? Come by at 18:30 for a short info session where we explain the roles and what being on the board involves.
After the session, we will have a speed date round where you can briefly speak with current board members, ask questions, and get a quick feel for each position.
The research lecture starts shortly after the speed date round and will take place in the same room.
Research Lecture by Marie Corillon
StarTime: Finding Signal in High-Frequency Data
High-frequency data often hides the real predictive signal among many noisy, highly correlated lags. Standard approaches either drop lags unpredictably or rely on rigid, manually chosen groupings.
In this lecture, Marie Corillon introduces StarTime, a data-driven method that automatically discovers optimal temporal structures. Using temporal trees, the model groups similar lags together to capture the underlying signal while filtering out noise.
The talk will also demonstrate how StarTime produces accurate and interpretable forecasts, with applications ranging from stock volatility prediction to GDP nowcasting.