Course Offerings SUMMER TERM
BACHELOR'S, MASTER'S and PhD Level
One Bachelors' module (Bsc), one Masters' module (Msc) and one PhD module (PhD) are offered during the summer term.
MODULE (Bsc): FINANCIAL DATA SCIENCE
- 9 ECTS (one full module); Vertiefungsstudium
- Grading is based on Python problem sets. No "classical" sit-in exam; rather a real-world environment of solving real-world problems in a small team. The problem sets start very simple and get more advanced, requiring a skillful combination of lecture material, which ranges from financial-, analytical-, statistical-, to data engineering concepts. ... Get together with your 2-3 best friends and join the ride.
- This terms' Python problem sets include: (i) strategic asset allocation with numerical optimization, (ii) extracting predictive features for equity return predictions from option markets, (iii) finding market-neutral (long-short) portfolio strategies with high and significant alphas, (iv) learning classical machine learning models and evaluate their statistical and financial performance. We likely extend the problem sets by two applications: assessing the real-time intraday news impact on asset prices and robust reinforcement learning strategies for optimal asset allocation.
- Lecture: Monday: 8:00am - 11:15am, Engler-Bunte Hörsaal, R40.50
- Tutorial (Python "Focus Sessions"): Monday: 11:30am - 1pm, Egon-Eiermann Hörsaal.
- First meet-up is on April 28th (due to Ostermontag on April 21st).
The topics are structured as follows:
1. MBA Intuition for Core Finance Concepts
- Case Study: Theory of valuation
- Case Study: Theory of optimal asset allocation
- Theory for determining the cost of capital,
- Running predictive regressions and linear factor model regressions
2. Python Fundamentals; applied to Mean-Variance Portfolio Choice and Factor Regressions
- Regression analyses for linear factor models
- Constrained optimization for portfolio construction and likelihood-based learning
- Introduction to key libraries (e.g., NumPy, Pandas, ...)
3. A Bird's Eye View onto Tradeable Financial Data
- Machine Learning: Why Finance Data is Different
- A trader's view on data: spot vs futures markets; trading rates vs cash-flows vs volatility
- Arbitrage in spot-, futures-, derivatives markets
4. Extracting Predictive Features for Stocks from Stock Options
- ECB case study on COVID-19
5. Risk-Neutral PDFs: A microscopic view into the future
- Introduction to the Chair's big data database
6. Artificial Neural Nets (ANN)
- Concept, training, validation, testing
- Application: learning the model that we used to generate prices of calls and puts
- Application: predicting next day's option prices
- Python: tensorflow, keras, sklearn
7. What Traders Know about the Black-Scholes-Merton Option Pricing Model
- Replication, martingale pricing
- Reasons for vol skew
- The Greeks
- P&L of option trading
8. Risk Management
- Scenario analysis, Value-at-Risk
- Time-series risk forecasting using GARCH and Stochastic Volatility Models
- Python: MLE for GARCH
- Python:; Kalman Filter MLE for Stochastic Volatility Models
9. Machine Learning Portfolios
- Discussion of the tapastry of machine learning modeling approaches in literature
- Critical assessment of their empirical performance
- From kitchen-sink predictive models for stocks and options to constrained deep learning approaches to end-to-end model-based reinforcement learning approaches
Dedicated Tutorials ("Focus Sessions"), offer insights into the workings of Python for big datasets. Topics include
(i) numerical optimization: constrained mean-variance efficient portfolios, empirical likelihood functions,
(ii) empirical asset pricing: constructing predictive features, efficient portfolio sorts for finding high alpha portfolio strategies,
(iii) supervised learning: predicting stock returns with machine learning methods, learning option pricing models,
(iv) intraday price variations: on the importance of economic events,
(v) model-based reinforcement learning: end-to-end robust portfolio decision making systems.
MODULE (Msc): ADVANCED MACHINE LEARNING AND DATA SCIENCE
- Topics: End-to-End model-based reinforcement learning, deep supervised learning, natural language processing: with applications to financial market data
- Kick-off: Discussion and Allocation to topics takes place first week of the lecture period. For a later start contact us. Also note; earlier start possible if you register online: you can use either https://portal.wiwi.kit.edu/forms/form/AdvML or https://portal.wiwi.kit.edu/forms/form/811
- contact: maxim ulrich ∂does-not-exist.kit edu
- A recent course review of a KIT computer science student: "Der Kurs <Advanced Machine Learning and Data Science> hat mir eine gute Möglichkeit geboten, auch ohne einen Finance-Hintergrund mein Wissen aus der Informatik erfolgreich anzuwenden. Zwar bekommt man jederzeit Hilfe bei Fragen, doch es war wirklich notwendig, sich intensiv mit dem Thema auseinanderzusetzen und eigene Ideen einzubringen, um den Kurs sehr erfolgreich abzuschließen. Besonders positiv fand ich, dass das Projektthema in Absprache mit den Veranstaltern nach meinen eigenen Fähigkeiten und Interessen gestaltet wurde, was mir viel Mitspracherecht und Flexibilität gab.
Insgesamt hat mir der Kurs als Informatikstudent die Chance geboten, mich auf das Thema Data Science im Finanzbereich vorzubereiten und wertvolle, berufsqualifizierende Erfahrungen zu sammeln." (Name of student is kept private).
MODULE (PhD): DEEP FINANCIAL ECONOMICS
- This module is a research oriented discussion group to assess the quality and impact of newly published research papers. Roughly 3-6 papers are presented per week and discussed.
- Researchers from other research groups and departments are very welcome to participate
- contact: maxim ulrich ∂does-not-exist.kit edu