PhD Defence by PhD Peter Nystrup: Dynamic Asset Allocation - Identifying Regime Shifts in Financial Time Series to Build Robust Portfolios

PhD Defence by PhD student Peter Nystrup, Wednesday, February 21, at 13:00 in Building 101A, Room S01

The purpose of the project was to develop and test dynamic asset allocation strategies based on identifying regime shifts in financial time series using models with time-varying parameters. The project has shown that dynamic asset allocation is key to avoiding the large drawdowns that characterize traditional, static investment portfolios during crises in financial markets. The 2008 financial crisis clearly showed that diversification is not sufficient to avoid large drawdowns, because correlations between risky assets tend to strengthen during times of crisis.
Through better modeling of the dynamics of financial return series - taking into account nonstationarity and abrupt changes - combined with a dynamic optimization approach, the project has shown that it is possible to control drawdowns. Part of the research builds on well-known engineering methods, including adaptive estimation and control-based approaches to multiperiod optimization, that have been adapted and successfully applied to financial data. The project has also led to development of new approaches to portfolio optimization, trade planning, as well as optimization-based approaches to segmentation of time series that have been made available as open-source software. These tools have several potential applications within both research and industry, for example when implementing and evaluating new trading models.
The project was a collaboration between Sampension, Technical University of Denmark, and Lund University supported by Innovation Fund Denmark. In addition, it included research collaboration with Stanford University.

Read more about this thesis in DTU Orbit.

Principal supervisor: Professor Henrik Madsen, DTU Compute
Co-supervisor: Professor Erik Lindström, Lund University
Co-supervisor: Head of Investment Analysis Bo William Hansen, Sampension
Co-supervisor: CIO Henrik Olejasz Larsen, Sampension

Professor Pierre Pinson, DTU Elektro (Chairman)
Professor Magnus Dahlquist, Stockholm School of Economics
Professor Lasse Heje Pedersen, CBS

Moderator: Niels Kjølstad Poulsen

After the defence there will be a small reception in building 303B room 128

Everyone is welcome!


ons 21 feb 18


DTU Compute



DTU, Building 101A, Room S01