the intelligent choice Excerpt plus FASTVaR long term risk + daily VaR from a consistent modelling framework Asset managers are under increasing pressure from clients and regulators to monitor FASTVaR with EMA s Excerpt market risk analysis system is a comprehensive solution and control the risks in their portfolios. Post Lehman that means estimating short term for equity, debt and multi-asset managers to assist in matching risk to expected return, VaR as well as understanding longer term risk exposures. EMA s FASTVaR is unique communicating your strategy to clients and satisfying regulators that you are risk in applying GARCH to the daily returns of a long term factor model, incorporating long aware. term attribution and daily VaR forecasts into a single framework. EM Applications - HOME EM Algorithm CLICK AND HOLD TO VIEW Copyright EM Applications 2011
FASTVaR FASTVaR: Unique multi-factor GARCH system EMA s Excerpt with FASTVaR is a proprietary combination of a robust statistical model that accommodates user-defined named factors with a daily GARCH derived multi-dimensional volatility forecast. Together they provide meaningful risk attribution plus both the stability of a longer term risk estimate and the responsiveness of a very short term risk forecast. FASTVaR builds on EMA s 20-factor long term statistical risk model to deliver fat-tailed, short term risk forecasts. Having generated a six year daily returns set for the long-term factors, a GARCH process is applied to project forward, producing a dynamic short term volatility prediction. By comparison with an exponentially weighted moving average (EWMA) approach, the use of GARCH allows for a reactivity term which takes account of the tendency of volatility to return to its long term trend. In addition, a skew-t distribution is incorporated to take account of the fat-tailed distribution of asset returns. FASTVaR backtest equally weighted portfolio of top 5000 global stocks, daily rebalanced. Fat-tail VaR FASTVaR risk forecasts apply skew and kurtosis parameters estimated from a skew-t distribution and use pricing models to take account of derivatives nonlinear returns to produce realistic VaR analyses. For reliable VaR forecasts through extreme volatility events. I think the EM algorithm is the best way to do factor analysis in the financial context and, having reviewed a number of implementations of EM, I concluded that EMA s was the most robust and efficient. Head of ALM and Quantitative Analysis, asset management division EM Applications - FASTVaR Copyright EM Applications 2011
COMPARED The advance of risk models Analogous to the chess queen or the completed puzzle, EMA s Excerpt with FASTVaR draws together the features of several fundamental parts into a single comprehensive whole. The earliest factor models were based on pre-defined betas. These were unable to represent systematic effects that were unrelated to available data and hence there developed statistical models which derived factors from the observed relationships between assets. However, as the factors were unnamed this approach was less useful for attribution. Separately a covariance based approach was developed to generate daily VaR estimates for multi-asset portfolios. Taking the best of each of these approaches EMA uses a statistical method to build a set of dimensional factors that have high explanatory power on which are stood a set of language factors that relate to familiar asset characteristics. Finally, by applying GARCH to forecast the daily volatility of the statistical factors, the EMA system brings the best of all three types of model together into a single package based on a consistent theoretical framework Fundamental model Multiple regression Long term stability System selected attribution Covariance model Exponentially weighted moving average Short term responsive Limited attribution capability Statistical model Principal components Long term stability User selected attribution EMA Excerpt + FASTVaR EM Algorithm + GARCH Long term stability & Short term responsive User selected attribution EM Applications - Compared Copyright EM Applications 2011
RISK ATTRIBUTION Fund manager friendly EMA s Excerpt portfolio risk analysis system has developed over 10 years to meet requests from a number of fund managers, so that the reports it produces have been tailored to satisfy a wide range of approaches to managing money. Recognising that the language used to describe investment bets is specific to each manager, the system allows the use of the fund manager s own factors for risk attribution. These factors can represent investment styles Asset Class Exposure This button will open an example of an Excerpt graph that shows the allocation of risk exposure between asset classes. such as value or growth, or can be categorisation schemes such as country or sector. Meaningful attribution Establishing the macro-economic forces and style drivers that affect portfolio returns requires an analysis of the relationships between assets that distinguishes between sensitivity to common effects and asset specific behaviour. Without this distinction many apparent relationships will be spurious. The EM algorithm is an interactive process that is specifically designed for finding the common factors behind a dataset of relationships. It produces a statistically robust model and consequently generates relevant and meaningful attribution. The ability to attribute risk to a set of standardised investment themes that we designed plus the combination of short and long term risk forecasts in one model was key to our decision to use EMA. Head of Risk, consultancy firm Comprehensive asset coverage The system is designed to accommodate all the assets typically used by a UCITS fund; essentially this is all liquid securities, related derivatives and composite assets built from these including: Equities; 50 country and regional models including exotic emerging markets Fixed income; including governments, corporates, swaps and forwards Composites; including mutual funds and ETFs Commodities, Property and Private Equity; can be represented by indices or traded funds Risk Decomposition This button will open an example of an Excerpt graph that shows the split between top-down and bottom-up bets and the key areas where risk is being taken. EM Applications - Risk Attribution Copyright EM Applications 2011
RISK CONTROL EMA s Excerpt is a user-friendly tool that helps us ensure that our managers build portfolios consistent with our portfolio design parameters without putting them in a straightjacket and while allowing us the freedom to determine exactly what we wish to control. Head of Front Office Risk, wealth manager Short and long term risk measures Longer term risk measures, typically based on 3 to 5 years of data, are useful for determining intrinsic risk, so that changes in portfolio risk levels are more likely to be a function of trading activity than market behaviour. Short term risk measures, with half-lives of around 10 days, are well suited to giving an instant view of current risk. EMA s systems are unique in providing both measures built around a consistent modelling framework. Easy to integrate There is a need for both high-level risk reporting for oversight purposes and for more detailed analysis available at the risk or fund manager s workstation. EMA s Excerpt produces reports as self-contained Excel workbooks VaR Histogram that recalculate when planned trades are entered. With such an intuitive framework for pre-trade analysis it becomes a simple matter to integrate risk awareness into the investment process. Scaleable While the desktop version of Excerpt can be installed and running within one day, there is also an API (Application Programming Interface) option which permits integration into larger systems for unattended bulk processing. Regulatory compliance Global investment fund regulation is driven by the EU in the form of the UCITS rules for public funds and AIFM for professional funds. Both formalise the responsibility of investment managers to take account of risk when seeking alpha, encouraging the integration of risk awareness into the investment process. The minimum requirement is that portfolio volatility or VaR should be monitored; while being able to report on both longer term intrinsic risk and very short term VaR, and attributing risk to the investment bets taken, is seen as best practice and a competitive advantage with clients. EM Applications - Risk Control Copyright EM Applications 2011
SOLUTIONS Solutions Example Excerpt Reports EM Applications has four core products, all built on EMA s proprietary implementation of the EM algorithm, designed to support investment professionals in the efficient and risk-aware management of investment portfolios. 1 Excerpt An easy-to-use portfolio risk analysis program. It produces a very wide range of analytical reports suitable for the risk management of single or multiple portfolios, invested in equities, bonds, derivatives Marginals/Top 10 Table Which of the holdings contribute most to risk at the margin and most to total risk? 2 3 4 or multi-asset. FASTVaR An adjustment to EMA s standard long term risk model that produces VaR estimates over short-term horizons, typically from 1 day to 1 month. Optema A quadratic optimiser that is a powerful, yet simple, tool. Designed to assist in idea generation and portfolio construction. Estema The EMA Factor Analysis Toolbox is a suite of routines that implements Stroyny s research in EM factor analysis, including the Combined Linear Factor Model and Robust extensions. Useful for building EM based factor models for risk and alpha generation purposes. Risk Allocation Report What is a portfolio s sensitivity to market, country, sector and style bets? Theme Betas Table What macro-economic factors (e.g. oil price, interest rate shocks) is a portfolio sensitive to? EM Applications - Solutions Copyright EM Applications 2011
ABOUT US EM Applications For over 10 years EM Applications has supported investment managers around the world with portfolio risk analytics that have enabled them to measure the risk of their portfolios, attribute the risk to its underlying causes and communicate their investment style and strategy to their clients. The risk factor model is built using a proprietary application of the EM algorithm which has proven its ability to accurately distinguish between systematic and asset specific risk and establish the set of systematic factors that best explain the patterns of relative asset returns. EMA s solutions are used by large, medium and small asset management firms for risk analysis and to assist with portfolio construction and alpha development. We have used EMA for many years and find it useful in helping us establish the main market themes to which our portfolios are exposed. We have experience of seeing the portfolios perform as predicted when certain macro-economic scenarios pan out. Head of Risk, asset management company The Three R s EMA s systems are designed to assist with the essentials of success in fund management. Risk:Return trade-off Providing insight into risks taken so they can be matched to expected returns. Raising assets Demonstrating mandate conformity and risk awareness to build confidence with clients. Regulatory compliance Meeting ever more demanding standards. EM Applications - About Us Copyright EM Applications 2011