Yevhen Havrylenko

I am a DAAD PRIME Fellow (https://www.daad.de/en/studying-in-germany/scholarships/daad-funding-programmes/prime/) and conduct research in asset-liability management, decision-making theory, and data science with applications to insurance and finance.

I conducted research on the optimal investment-consumption strategies and on interpretable machine learning with applications to insurance.

I conducted 5 research project that resulted scientific publications publications and did 3 applied projects for ERGO & Munich Re (see section below). In addition, I supervised 3 Master theses, 3 seminar works and the TUM-ERGO Machine Learning team of 5 students. Finally, I co-organized 3 academic events, including the International Congress on Insurance: Mathematics and Economics 2019 in Munich.

Department “Global Property & Casualty Actuarial Pricing” Created a tool that recommends the next-best pairs of interacting variables for generalised model models (GLMs) Methodology is based on my research paper “Detection of interacting variables for GLMs via neural networks” co-authored with J. Heger Implementation: R (frontend) and Python (backend, mainly keras and tensorflow packages) Prepared technical as well as methodological documentation of the developed tool

Departmet "Strategic Asset Allocation". Created a tool for robust computation & analysis of relevant investment portfolios risks such as Market Value-at-Risk (VaR), Credit VaR, Foreign-Exchange VaR, Asset-Liability Mismatch Risk (ALMR). Tool functionality: total risk factor VaR & ALMR, disaggregated risks (e.g., for asset classes), marginal risks, robustification using L-estimators. Implementation: MATLAB (computation & plotting) and Excel (user interface).

Department "Strategic Asset Allocation". Analyzed algorithms suitable for clustering financial assets. Conducted clustering analysis of the asset universe of ERGO Group. Created a user-friendly Python-Excel clustering tool for asset allocation decision support.

Course in Financial Market Volatility. Key topics: - symmetric and asymmetric GARCH, CCC- and DCC-GARCH, orthogonal GARCH models - stochastic and local volatility models - volatility trading

Course in Statistics for Business Administration

Project centred around IRBA audit preparation of a German bank. Analyzed deviations between internal models and their prototypes to control credit risk. Corrected found errors using PL/SQL. Developed further the credit decision process prototype in PL/SQL.

Team "Credit Portfolio Risk Measurement & Methodology". Analyzed performance of algorithms for stochastic LGDs generation in C++. Developed and implemented in C++ an efficient algorithm for stochastic LGDs generation. Calibrated a structural credit portfolio model using R.

In my dissertation, I study the optimal investment and risk-sharing strategies for decision makers with constraints, e.g., insurance companies. I solve the corresponding portfolio-optimization problems by advancing existing methodologies to tackle the presence of risk-sharing mechanisms as well as constraints on terminal wealth or investment strategies. Link to dissertation: https://mediatum.ub.tum.de/?id=1692226

Mathematical finance: stochastic analysis, continuous time finance, fixed income markets, portfolio analysis. Statistics: time series analysis, computational statistics, generalized linear models. Optimization: nonlinear optimization advanced, optimization methods in machine learning

Foundations: algebra, mathematical analysis, discrete mathematics, probability theory, statistics. Applied mathematics: data analysis, operations research, numerical methods, decision making theory, econometrics. Computer science: programming (C++, C#), algorithms