Speaker: Prof. Emeritus Günter Schmidt, Saarland University, Germany
Title: Online Algorithms in Finance
Date: 2026-04-23
Time: 15:00 - 16:30 (EEST)
Location: (Hybrid) FORTH Central Building, Orphanoudakis' Room + Teams
Host: G. Tzagkarakis and G. Tsagkatakis, FORTH-ICS
Teams Link
Abstract:
The talk introduces how online algorithms provide a rigorous uncertainty framework for trading, portfolio management, and risk control. The focus is on trading problems from the perspective of quantitatively minded practitioners and researchers who analyze performance guarantees added by backtests. First we concentrate on core concepts of online algorithms for sequential decision-making with partial information. These results are applied to financial problems such as asset allocation and conversion. Classic examples like online portfolio asset selection and secretary-type models for conversion illustrate how one can prove worst-case bounds on such algorithms. Then we survey online-algorithmic techniques combined with risk estimations usable in finance. Topics include online gradient methods and parameter tuning. From these techniques heuristics for practical applications are derived. Decision variables are transaction prices and volumes using worst case analysis and Kelly criteria. The talk highlights the tension between theoretical guarantees and implementation realities. It finishes with open challenges.
Short bio: Günter Schmidt is an Emeritus Professor at Saarland University in Saarbrücken and a Senior Professor and Head of the Research Group “Operations Research” at the German University of Digital Science in Potsdam. His research interests include combinatorial optimization, with a focus on theoretical advances in solving scheduling and trading problems, as well as practical developments in game design.