Abstract

Financial markets have always been informed and transformed by a wide range of statistical and computing methodologies. I will explore the critical role that high-dimensional time series, extreme value analysis, hierarchical modeling, simulation, and stochastic processes continue to play even in this dynamic era of AI. Examples will include multivariate time series methods to understand the changing co-volatility in markets during stable and extreme conditions, approximate Bayesian computation to model FX prices using an alpha-stable distribution, and stochastic models that illustrate how news affects futures prices differently and its varied impact on market volatility on the upside and downside. While learning algorithms can provide similar insights, it is not necessarily a better path to the answers.