Improving #Portfolio #RiskManagement with ICA
Independent Component Analysis (ICA) is a powerful modern signal processing technique. The key idea of ICA is to linearly map the observed multivariate time series into a new space of statistically independent components (ICs).
Modern risk management theory works on the concept that given some returns data, a few assumptions about dependence and a family of probability distributions, risk can be modeled effectively. Being a novel technique as it is, ICA can jointly estimate a set of statistically independent factors and their loadings. What makes ICA different from conventional approach is that it escapes the need to make heavy assumptions about probability laws.
Dr. Andrew Kumiega and Greg Sterijevski are doing cutting-edge research on ICA and its practical application. Andy is a professor at IIT and Director at Infinium Capital Management, and Greg a risk system expert and Director at UBS Oâ€™Connor. Together, they will present their research on Portfolio Risk Management with ICA: An illustration and comparison to parametric Value at Risk (VaR) and Expected Shortfall (ES) at the Quant Invest Chicago 2012 conference. On June 25, they will lead a hands-on, interactive and comprehensive workshop to teach everything about ICA, including using some third party hedge fund return data to estimate risk parameters for a set of archetypical portfolios.
(Research description is provided by the authors.)