Option Implied Dependence (FWOAL942)

Options prices provide information on the future probability distribution of asset returns. However, information on the dependence among assets is difficult to extract and interpret and there are no model-free methods to obtain the full joint distribution among assets. The Chicago Board Options Exchange publishes option implied correlation indices since 2009. However, their methodology suffers from several limitations. The first objective is to propose a new method to extract the most plausible joint distribution among the assets (i.e., the maximum entropy distribution) that is consistent with all options prices. The information is extracted daily and makes it possible to provide a dynamic of the dependence over time and, ultimately to select the best dynamic models consistent with that information. My second objective is to apply this new
approach to systemic risk assessment and to develop a new indicator for early detection of systemic risk and financial crises. I will also present novel arbitrage strategies and improved investment
strategies. As a third objective, I will assess model risk on the first two objectives. Specifically, I will use tools from operational research and machine learning to assessing the range of possible distributions that are consistent with options prices as well as to evaluating where the implied maximum entropy distribution stands in this range. I will also assess the uncertainty on systemic risk assessment and arbitrage detection strategies

Keywords:Financial options, dependence modelling

Disciplines:Econometric modelling