Vortrag von Hanna Wutte
Datum: 30.11.23 Zeit: 12.15 - 13.45 Raum: Y27H12Abstract: Dynamic decision making in mathematical finance typically arises in form of trading in a financial market, often including very general constraints (e.g., on liquidity and trading volumes) for purposes of hedging, pricing or portfolio optimization. Neural networks are increasingly leveraged in that optimization via parametrizations of trading strategies that can easily incorporate common market frictions. In order to choose an appropriate deep learning algorithm from a large pool of existing approaches, it is indispensable to understand their assumptions and the implications of their use. We contrast common approaches to deep dynamic decision making, in particular with a view to portfolio optimization in energy markets.