PRE-CONFERENCE WORKSHOP 1: Investment risk: Risk factor models and beyond
Speaker: Grant Wang, Senior Vice President, Head of Research, HIGHSTREET ASSET MANAGEMENT
This workshop covers quantitative investment risk analysis, concentrating on major asset classes such as stocks and fixed income. The analysis relies on statistical modelling of investment portfolio returns, complemented by economic, financial and behavioral perspectives. The multidisciplinary approach is crucial in today's ever changing environment of investment risk. In the risk community, worst-case losses analysis represented by VaR is overwhelming. The variance analysis focused by this workshop will provide complementary and new insights on investment risk.
8:30 Registration and breakfast
9:00 A synthesis on risk factor models
- Under normal distribution of portfolio returns, the investment risk can be completely characterized by return variance
- Variance of portfolio return can be specified by a factor model
- There are three types of equity risk factor models widely accepted by industry:
- Fundamental is the most utilized and can attribute return and risk (from both absolute and active aspects) to style and other fundamental factors, sectors/industries, and countries.
- Macroeconomic can be used to understand the portfolio's exposures to macroeconomic factors: interest rate, oil price, etc.
- Statistical factors are purely derived from equity returns particularly over a short investment horizon
- There are two types of credit risk factor models: structural and reduced form
- Reduced-form model has become the standard because it provides tractable formulas for analyzing credit-sensitive assets. The model is based on an exogenously specified conditional rate of default or intensity.
- Structural or cause-and-effect model is based on a specified definition of default
10:30 Morning coffee break
11:00 Risk factor models in action
- The most important application of equity risk model is portfolio construction. In particular, the risk model provides risk estimates that are inputs to the objective function or portfolio constraints. Moreover, to achieve higher risk adjusted returns, portfolio managers may want to have dynamic portfolio constraints which are more adaptive to market environments. The risk model forecasting can be an indicator of specific market environment -high or low risk in the market.
- Another usage of risk model is to derive the implied alphas of stocks from the portfolio holdings. This can be conducted by reverse optimization. Some portfolio managers may not (want to) have an alpha model but do want to check the implied alphas and use the information to refine the portfolio. This is an iterative process (initial portfolio to implied alphas to modified portfolio).
- Factors in risk model can be utilized as proxies of alpha factors. Note that the risk factors are usually more generic as they are defined by the vendor while the alpha factors are customized and intend to be used specifically for the chosen investment strategy. The performances of the factors, and their interdependencies, when compared with historical scenarios can provide the context of current market behavior. For example, the correlation between a factor and some valuation factor, compared with historical average, indicates whether the factor is relatively cheap to invest in or not.
1:30 Investment risk goes by regimes
- Risk regimes change often suddenly and in unexpected ways. To prepare adequately and adjust quickly, it requires deep understanding of the assumptions and vulnerabilities of various risk factor models
- There are different ways to identify the regimes. For example, gaps between fundamental and statistical models' risk estimates may signal disruptive market events and regime changes
- In the elevated risk regimes, it is more important to hedge downside risk
3:00 Afternoon coffee break
3:30 Case study - autopsy of risk of momentum investing
- Momentum investing buys past winners and selling past losers. It is pervasive over multiple time periods and across numerous asset classes
- Momentum doesn't always work. During this period, it may incurs extreme risk and losses
- Risk factor models can attribute momentum's risk to important risk factors such as market beta and help portfolio managers position bets in anticipation of market movement
- Risk regime analysis can help avoid potentially large losses of momentum strategies
5:00 End of workshop
View the other workshop agendas:
- PRE-CONFERENCE WORKSHOP 2, October 20: Central Clearing and Central Counterparties
- POST-CONFERENCE WORKSHOP 3, October 23: Quantifying the risk of incremental model changes
- POST-CONFERENCE WORKSHOP 4, October 23: Exchange Traded Funds: Price dynamics, tracking errors, trading strategies, and option pricing
- *NEW* Risk Behavioral Finance Forum, October 23