By Attilio Meucci
Bloomberg ALPHA, Portfolio Analytics and Risk
Abstract
We introduce a unified, modular framework to build a risk platform that properly estimates, represents, hedges and manages risk. To do so, we analyze and fix common misconceptions on linear factor modeling that translate into incorrect risk numbers and flawed optimized portfolios.
We explore applications stemming from this modular approach: the joint use of random matrix theory for risk estimation and of fundamental factor models for risk interpretation to optimally cater to both risk- and portfolio-managers; the correct implementation of linear factor models and optimal hedging for options not based on Greeks; a global equity model that is fully consistent with more granular regional equity models; a consistent approach to performance and risk attribution; point-in-time style analysis, instead of the standard trailing regression approach; portfolio-based risk attribution, to best express a portfolio in terms of a series of trades; dynamic factor pools, to extract the best few explanatory factors at each point in time.
Keywords: Regression, Estimation, Interpretation, Copula, Marginal Distribution, Random Matrix Theory, Style Analysis, Optimal Hedging, Selection Heuristics, GICS Industry Classification, Monte Carlo
Fully commented code supporting the above case studies is available at MATLAB Central File Exchange.
Finance Focus
MoneyScience Twitter
The European Business School London MBA programme provides students with the skills and experience to undertake the challenges of senior-level management. The programme has an international focus, underpinned by an integral 6-10 week Study Period Abroad which can be spent taking MBA modules at a leading university or business school in Brazil, Belgium, Canada, Denmark, South Africa or Spain. We offer two start dates a year in January and September.