Statistical Learning and Digitisation in Financial Mathematics and Actuarial Science (SDFA)

The advancing digitalisation in the insurance and financial market includes opportunities and risks that we would like to model, predict and control from a mathematical and statistical perspective in the research field SDFA. 

The research field combines stochastic modelling and statistics of mathematical finance with methods of machine learning, specifically statistical learning. This field is concerned with the recognition and modelling of correlations in large amounts of data, and predictive analyses are also carried out.

Influencing factors are considered, such as:

  • Data volume – which model can be used for which data volume?
  • Data quality – which data sets can be included in the models? What kind of dependencies exist?
  • Parameter uncertainties – how sensitive is the model and how can calibration be performed?
  • Explainability – which methods and models can be developed that not only show correlations, but can also analyse and interpret them?


An interplay of methods from statistics, stochastics, optimisation and numerics brings solutions to questions of portfolio optimisation and asset allocation, risk management and actuarial issues.


Current projects

  • Modelling of credit spreads through  regime-switching in neural networks
  • Sustainable Finance: ESG ratings in portfolio optimisation
  • Forecasting electricity spot prices in switching LSTMs
  • Regimes and sentiments in crypto currency markets
  • Robustifying Markowitz
  • Robustified Markowitz approach for cryptocurrencies
  • ESG and Credit ratings: dynamics for various market regimes 
  • Portfolio Diversification based on Risk Profile Clustering
  • COVaR portfolios with digital assets
  • Modelling Term Structure
  • Bermuda-Swaption
  • Implied Probability Densities