Long-Short-Term Memory Neural Networks in Varying Regimes

In-person and Virtual Participation Available

Machine learning (ML) models are now widely applied to understand and predict patterns in financial markets. Classification trees and artificial neural networks are popular choices for modelling market behavior and risk measures. ML models are powerful tools, but must be handled with care to produce interpretable and reliable results for risk managers. 

As data-driven decision processes made their way into risk management, estimation and model identification issues have arisen. These might result from too much or too little data and be reinforced by model choices that are inappropriate given the prevailing market conditions.

To capture varying market conditions, we develop a model combining a long-short-term memory neural network (LSTM) with a Hidden Markov model (HMM).

The presentation will cover:

• Detection of hidden states in observed financial time series

• Incorporation of regime information into LSTM

• Forecasting of corporate credit spreads in changing market regimes

• Accuracy of neural network and HMM-LSTM prediction applied to corporate credit spreads of three European countries

Furthermore, an outlook is given on an HMM-LSTM ensemble model, where regime-switching information works as a gating function to activate a neural network. We apply this to time series forecasting in electricity markets. 

Agenda

6:00 – 6:30pm: Registration

6:30 – 7:30pm: Presentation

7:30 – 9:00pm: Networking Reception

Meeting Location: 

Deutsche Bank AG

Otto-Suhr-Allee 6-16, 10585 Berlin

Online link to follow

Details

September 28, 2022
6:00 PM - 9:00 PM

Hybrid

Join Online
_{Link Emailed Upon Registration}

or

Deutsche Bank AGOtto-Suhr-Allee 6-16, 10585 Berlin, Otto-Suhr-Allee 6-16,, Berlin, 10585