Machine Learning in Finance Spring 2022

Lecturer
Prof. Dr. Josef Teichmann
Coordinator
Songyan Hou

Content

This course will deal with the following topics with rigorous proofs and many coding excursions: Universal approximation theorems, Stochastic gradient Descent, Deep networks and wavelet analysis, Deep Hedging, Deep calibration, Different network architectures, Reservoir Computing, Time series analysis by machine learning, Reinforcement learning, generative adversersial networks, Economic games.

Prerequisites

Bachelor in mathematics, physics, economics or computer science.

Lectures

Lectures take place on Mon 10-12 at HG G 5 and Wed 11-12 at HG G 3 .

Lectures and classes will not take place during Easter week from Friday, April 15 until Sunday, April 24.

Teaching is currently planned to take place in person, although this may change at any point depending on the evolution of the pandemic and the measures taken by ETH Zürich. But live stream of the lectures are available here .

Lecture Notes

Lecture notes are provided as ipython notebooks or in form of slides as well as of classical notes.

Exercise classes

Exercises will be available in the exercise class. Students are expected to voluntarily do calculations and present results in class. Solutions will also be released right during the exercise class.

Exercise classes take place on Wed 10-11 at HG E 21 , LFW C5 .
Zoom link for tablet sharing in LWF C5

Exercise classExercise sheetReferences
Wed 23 Feb. Exercise sheet 1
Exercise notebook 1
Solution sheet 1
Solution code 1
The Faber–Schauder system
Wed 2 Mar. Exercise sheet 2
Exercise notebook 2
Solution sheet 2
Solution code 2
Interpolation and approximation by polynomials (Chapter 6)
Lebesgue’s Proof of Weierstrass’ Theorem
Bernstein polynomials and brownian motion
Wed 9 Mar. Exercise sheet 3
Solution sheet 3
Ex class recording 3
Neural ordinary differential equations
A theoretical framework for backpropagation
Wed 16 Mar. Exercise sheet 4
Solution sheet 4
How backpropagation works
Wed 23 Mar. Exercise notebook 5 Deep Hedging
PFhedge
Wed 30 Mar. Exercise sheet 6
Solution sheet 6
Deep Hedging under Rough Volatility
Stochastic Finance: An Introduction in Discrete Time
Wed 6 Apr. Exercise sheet 7 Calibration of local stochastic volatility models to market smiles: A monte-carlo approach
Wed 13 Apr. Deep Calibration of Heston model
Wed 20 Apr. (Easter Break)
Wed 27 Apr. Exercise sheet 8 Differential equations driven by rough paths
Wed 4 May. Exercise sheet 9 A primer on the signature method in machine learning
Wed 11 May. Exercise sheet 10 Applications of Signature Methods to Market Anomaly Detection
Discrete-time signatures and randomness in reservoir computing
Wed 18 May. Exercise sheet 11 Mathematics of Financial Markets
Reinforcement learning: An introduction
Wed 25 May. Exercise sheet 12 Valuing american options by simulation: a simple least-squares approach
Optimal stopping via randomized neural networks.
Wed 1 Jun.(LFW C5) Exercise notebook 12 Valuing american options by simulation: a simple least-squares approach
Optimal stopping via randomized neural networks.

Literature