Topics

Collection of articles on specific topics.

Probabilistic Time-Series Analysis and Forecasting

Many consider forecasting to be just a regression problem with lagged observations as features. Therefore, point forecasts are often the cookie cutter approach as they are in cross-sectional problems.

I tend to disagree with this notion and hence write quite a bit about the issues of point forecasts:

• When Point Forecasts Are Completely Useless
• Why I prefer Probabilistic Forecasts - Hitting Time Probabilities

GARCH-type models

Probably the most important classical time-series model for dealing with stock-price returns.

• Random Forests and Boosting for ARCH-like volatility forecasts
• Multivariate GARCH with Python and Tensorflow
• Let's make GARCH more flexible with Normalizing Flows
• Varying Coefficient GARCH

More Time-Series Analysis

Other thoughts on time-series problems

• Winning with Simple, not even Linear Time-Series Models