QFRG & DSLab seminar [10.10.2022]

We kindly invite you to join the 22nd seminar organised jointly by QFRG (Quantitative Finance Research Group) and DSLab (Data Science Lab).

During the meeting Paweł Sakowski, Ph.D. and Maciej Wysocki, student of the Doctoral School of Social Sciences will present the results of their research “Investment Portfolio Optimization Based on Modern Portfolio Theory and Deep Learning Models".

The study investigates the problem of an appropriate variance-covariance matrix estimation in the Modern Portfolio Theory. A novel approach combining the Markowitz framework with deep learning models was proposed and tested on portfolios of stocks and cryptocurrencies. Long short-term memory neural networks along with probabilistic deep learning models DeepVAR and GPVAR were compared with the classical approaches: frequentist estimator, exponentially weighted moving average, and shrinkage estimators. The LSTM model provided the best results across multiple investment schemes in terms of the information ratio, while the classical estimators in general performed the worst. Although the probabilistic deep learning models produced stable results, these strategies performed worse than the LSTM-based ones.

The meeting will be held on October 10th 2022 at 6:30 p.m. via Google Meet platform and will be held in English.

Link to the meeting: meet.google.com/gnr-gfbh-fxc