Vienna - Ivan Zhdankin, Douglas Vieira - Machine Learning in Finance

The Thalesians and Thales of Miletus invite to the first seminar in Vienna. Ivan Zhdankin and Douglas Vieira from JPMorgan and Imperial College London will hold talks about two topics at the core of modern finance.

Ivan Zhdankin: Key Aspects of Machine Learning in Finance

Abstract: The talk is about the fundamentals one should know when building machine learning models. The main issues discussed are overfitting, underfitting, bias and variance and how we approach those challenges, namely with cross-validation, bootstrap and regularization techniques. During the talk we will go through classic examples demonstrating those issues and will solve them by the listed techniques. At the end of the talk we will build a ML model going through every step of development: the model estimates the market impact in the crypto currency market.

Douglas Vieira: Microstructure of Option Prices: Reconciling Small and Long Time Dynamics

Abstract: Even though the option pricing literature is very rich, little is known about its dynamics at a microstructure level. In fact, the modelling of the microstructure of option price dynamics is a bit paradoxical. The price of an option reflects the expectation of its underlying asset price and volatility over a large time scale. Hence, while studying the microstructure of option, the global structure of its underlying price process cannot be ignored. Using asymptotic arguments, we address this very paradox in two directions. First, we distinguish the features of the global 'fundamental' dynamics are preserved locally from those that are lost. Second, we add microstructure effects to this 'fundamental' price process and show that they vanish at a larger time scale. The main outcome of this analysis is to precisely identify in which form the option price depends on volatility at small time scales.