Machine Learning about Implementable Portfolios
Freitag, 25.11.22, 12:00-13:00, online: Zoom
We develop a framework that integrates trading-cost-aware portfolio optimization with machine learning (ML). While numerous studies use ML return forecasts to generate portfolios, their agnosticism toward trading costs leads to excessive reliance on eeting small-scale characteristics, resulting in poor net returns. We propose that investment strategies should be evaluated based on their implementable ecient frontier, and show that our method produces a superior frontier. The superior net-of-cost performance is achieved by integrating ML into the portfolio problem, learning directly about portfolio weights (rather than returns). Lastly, our model gives rise to a new measure of "economic feature importance".
What works best? Methods for ranking competing treatments
Freitag, 9.12.22, 12:00-13:00, online: Zoom
Systematic reviews often compare multiple interventions simultaneously. Data from such reviews form networks of interventions and are synthesized through network meta-analysis, a technique which is used to combine evidence coming from all possible paths within the network. The main output of network meta-analysis is the set of all relative effects between competing treatments. A treatment hierarchy is also often of interest and several ranking metrics exist. In this talk I will describe available methods for ranking treatments and a method we developed in order to attach ranking to a clinically relevant decision question. Our approach is a stepwise approach to express clinically relevant decision questions as hierarchy questions and quantify the uncertainty of the criteria that constitute them. I will demonstrate the approach using the R package nmarank, available in CRAN.
Valuation and Risk Management of Guaranteed Minimum Death Benefits (GMDB) by Randomization
Freitag, 13.1.23, 12:00-13:00, Hörsaal II, Albertstr. 23b
Randomization is a technique in Finance to replace known quantities (like the time to maturity) by random variables. This sometimes gives moments or quantiles of the payoff in closed-form, avoiding any kind of integration, Fourier inversion or simulation algorithm. We apply this idea to insurance and Guaranteed Minimum Death Benefits (GMDB) where payoff dates are per se random. The remaining lifetime is expanded in terms of a Laguerre series while the financial market follows a regime switching model with two-sided phase-type jumps. For European-type GMDBs, we obtain the density of the payoff in closed form as a Laurent series. Payoff distributions of contracts with path-dependent guarantee features can be expressed in terms of solutions of Sylvester equations (=matrix equations of the form AX + XB =C).\n\nThis is joint work with Griselda Deelstra (Université Libre de Bruxelles).\n\nA paper version is available here: Deelstra, Griselda and Hieber, Peter, Randomization and the Valuation of Guaranteed Minimum Death Benefits, https://ssrn.com/abstract=4115505.
A phase field model for soma-germline interactions in Drosophila oogenesis
Freitag, 20.1.23, 12:00-13:00, online: Zoom
In [1], we study the signals mediating the mechanical interaction between somatic epithelial cells and the germline of Drosophila. We discover that, during the development of the egg chamber, the transcriptational regulator “Eyes absent” (Eya) modulates the affinity of the apical surface of epithelial cells to the nurse cells and the oocyte in the egg chamber. Using a phase field model, we develop a quantitative, mechanical description of epithelial cell behavior and demonstrate that the spatio-temporal expression of Eya controls the epithelial cells’ shape and movement during all phases of Drosophila oogenesis to establish a suitable match between epithelial cells and germline cells. Further we show that differential expression of Eya in follicle cells also controls oocyte growth via cell-cell affinity.\n\n \n\n[1] V. Weichselberger, P. Dondl, A.-K. Classen (2022): Eya-controlled affinity between cell lineages drives tissue self-organization during Drosophila oogenesis. Nat Commun 13(1):6377. DOI: 10.1038/s41467-022-33845-1
Generalized Covariance Estimator
Freitag, 27.1.23, 12:00-13:00, Hörsaal II, Albertstr. 23b
We consider a class of semi-parametric dynamic models with iid errors, including the nonlinear mixed causal-noncausal Vector Autoregressive (VAR), Double-Autoregressive (DAR) and stochastic volatility models. To estimate the parameters characterizing the (nonlinear) serial dependence, we introduce a generic Generalized Covariance (GCov) estimator, which minimizes a residual-based multivariate portmanteau statistic. In comparison to the standard methods of moments, the GCov estimator has an interpretable objective function, circumvents the inversion of high-dimensional matrices, and achieves semi-parametric efficiency in one step. We derive the asymptotic properties of the GCov estimator and show its semi-parametric efficiency. We also prove that the associated residual-based portmanteau statistic is asymptotically chi-square distributed. The finite sample performance of the GCov estimator is illustrated in a simulation study. The estimator is then applied to a dynamic model of commodity futures.\nChristian Gourieroux & Joann Jasiak (2022): Generalized Covariance Estimator,Journal of Business & Economic Statistics, DOI:10.1080/07350015.2022.2120486