Purpose built multi-asset models
As the EMA models are built directly from observed market performance, without any attempt to assign securities to a given category or measure them by reference to any attribute, the approach naturally extends beyond equities – for which linear factor models were originally built – to any traded asset class. Not only does the system then estimate the volatilities and correlations within an asset class, but extends to cross-asset correlations between all securities and derivatives regardless of asset type, origin or level of complexity.
By incorporating pricing models for derivatives and by mapping corporate debt onto the systematic exposures of the underlying company, the EMA system provides a single solution for a mixed asset portfolio.
It is possible to consider the portfolio as a whole, estimating the proportion of risk attributable to investment themes appropriate to the nature of the portfolio. Alternatively, or in addition, the portfolio can be decomposed into asset class components and each examined separately.
This gives a comprehensive, centralised and consistent view of risk within one system. At times of market stress, when correlations between different asset classes tend to increase, the single model provides the appropriate solution.