Mathematical Finance & Financial Data Science Seminar
Learning Hidden Markov Models by Penalizing Jumps
Speaker: Peter Nystrup, Postdoctoral Fellow, Lund University & Visiting Researcher at NYU Courant
Location: 60 Fifth Avenue Room 527
Date: Wednesday, October 23, 2019, 3:30 p.m.
Synopsis:
Hidden Markov models are a popular choice for inferring the hidden state of financial markets. When a hidden Markov model is misspecified or misestimated, it often leads to unrealistically rapid switching dynamics. In many applications, however, the model is only useful if the underlying state sequence has a certain level of persistence. We propose a novel estimation approach based on clustering temporal features while penalizing jumps. We compare the approach to spectral clustering and the standard approach of maximizing the likelihood function in an extensive simulation study and an application of financial data. The advantages of the proposed jump estimator include that it learns the hidden state sequence and model parameters simultaneously and faster while providing control over the transition rate, it is less sensitive to initialization, it performs better when the number of states increases, and it is robust to misspecified conditional distributions. The value of estimating the true persistence of the state process is illustrated through a simple trading strategy where improved estimates result in much lower transaction costs.
Important: Non-NYU attendees need to email Daisy Calderon, mojar@cims.nyu.edu, by October 22 to be added to the guest list in order to get access to the building. Room size is limited and seats are available on a first-come first-serve basis.
Peter Nystrup is a Postdoctoral Fellow in the Division of Mathematical Statistics at Lund University in Sweden and in the Department of Applied Mathematics and Computer Science at the Technical University of Denmark. He is also the Head of Research at startup quant hedge fund Annox and currently a Visiting Postdoc at NYU Courant. He has previously been a Visiting Researcher at Stanford University and has worked in equity sales at Nordea Markets, in the investment department at Danish pension fund Sampension, and as an external consultant on advanced analytics at the energy company Ørsted.
Dr. Nystrup earned his B.Sc. in Engineering degree in Mathematics and Technology from the Technical University of Denmark (DTU) in 2012, followed by a M.Sc. (Hons.) in Engineering degree in Mathematical Modeling and Computation in 2014. In 2018, he was awarded the Ph.D. degree in Engineering from DTU upon completion of a research project on dynamic asset allocation and identification of regime shifts in financial time series. His research has been published in leading journals covering topics from quantitative finance and portfolio management to forecasting, optimization, and operations research.