Time Series (Analysis and Prediction)

• Analyzing time series models using (Variational) Autoencoders, Restricted Boltzmann Machines or Generative Adversarial Neural Networks
• Bayesian Statistics including Gaussian Process Regression incl. Multi-Output GP for Data Augmentation, Filling imputed TS, Data Cleansing or Outlier Detectio
• Signatures of Stochastic Processes

Model Free Hedging

• Model free hedging via Reinforcement Learning
• Calculating conditional expectations by kernel density estimation or Gaussian mean mixtures
• Hedging strategies based on generated signatures and conditional expectations based on purely data driven and model free methods

Data Cleansing / Anomaly Detection

• Use the models for combining the results with extreme value theory to anomaly detection
• Generate synthetic data having “learned” characteristics for:
  
  • Stress testing
  • Asset allocation
  • Hedging
• Use the methods for briding gaps in observed time series data