Seminarium 26.04.2022
dodano: 22/04/2022
przez: Dariusz Zawisza
  • Referent : Marcin Pitera
  • Tytuł referatu : Discrete-Time Risk Sensitive Portfolio Optimization with Proportional Transaction Costs
  • Termin: Tuesday 26.04.2022, 12:15, room 1093
  • Abstract: In this talk we discuss a discrete-time risk sensitive portfolio optimisation over a long time horizon with proportional transaction costs. After providing a gentle introduction into risk-sensitive portfolio optimisation, we show that within the log-return i.i.d. framework the solution to a suitable Bellman equation exists under minimal assumptions and can be used to characterise the optimal strategies for both risk-averse and risk-seeking cases. Moreover, using numerical examples, we show how a Bellman equation analysis can be used to construct or refine optimal trading strategies in the presence of transaction costs.

Seminarium 12.04.2022
dodano: 08/04/2022
przez: Dariusz Zawisza
  • Referent : Dariusz Zawisza
  • Tytuł referatu : Różne koncepcje pomiaru ryzyka
  • Termin: Wtorek 12.04.2022, 12:15, sala 1093

Seminarium 05.04.2022
dodano: 01/04/2022
przez: Dariusz Zawisza
  • Referent : Jarosław Duda
  • Tytuł referatu : Old and new statistical models in data compressors
  • Termin: Wtorek 05.03.2022, 12:15, sala 1093
  • Abstrakt: Data compression, widely used by our electronic devices, allows to reduce file sizes by exploiting their statistical dependencies. They often first transform the file into a sequence of symbols with estimated probability distributions (s_i, P_i), then entropy coding allows to encode this sequence into ~sum_i lg(1/P_i(s_i)) bits to be stored or transmitted. Hence, log-likelihood improvements of statistical models can be directly seen as bits/value savings in size of the compressed file. Such models should be computationally inexpensive for fast data processing. I will introduce to this topic and would like to discuss new promising approaches especially for genetic data and image compression (e.g. context binning, model clustering, adaptivity, exponential power distribution, sigma prediction, canonical correlation analysis).