### The Global Multi-Asset Risk Model

The “Global Multi-Asset” risk model contains risk factors that touch on all four asset classes: Equities, Fixed Income, Commodities, and FX/Currencies. That risk model is the default risk model used in the system, although there is a project currently in development that will allow the user to choose additional Risk Models.

All of the factors in the risk model are interdependent, so certain mathematical operations are performed in order to reduce collinearity and improve factor orthogonality. This factor interdependence also means that it is not just the factor construction that is important, but also decision of which factors will be included in the construction of the Risk Model.

Factors that reside in the Global Multi-Asset Risk Model are constructed in two different ways: Single Factors and Component Factors. These two methods are meant to be a balance of “investable” factor exposure and “pure” factor exposure.

Factors are collected from multiple sources

Bloomberg, FRED, S&P, Barclays, BNP Paribas, Credit Suisse, JP Morgan, Morgan Stanley, Nomura, Societe Generale, and public sources. The primary sources for style-driven factors are large global investment banks, while the primary sources for asset-class driven factors are public sources and other data providers.

Factors are sorted based on being primary or secondary factors.

This sort is not specifically empirical like the rest of the process, but is generally based on empirical data. For example, it is widely acknowledged that equity risk is a dominant risk factor for most portfolios, whereas hedge fund crowding is likely not a primary risk factor for most portfolios.

Single Asset Factors

Factors constructed using the Single Asset method are generally understood factors for US investors. For example, the “Equity” risk factor is constructed with a single equity index (the MSCI All Country World Index).

Component Factors

Component factors are constructed using multiple indices. The underlying indices are normalized, and then principal component analysis is performed on the index set in order to distill the most common characteristics of the factor set. The first principal component, the multi-dimensional mean, is then used to represent the factor.

### Volatility Scaling

Volatility scaling is performed both when creating the risk model (at the factor level) as well as when performing the factor analysis (at the asset level). After the analysis, the asset is re-scaled back to its original volatility. This normalization helps to ensure parameters like volatility do not corrupt the data. The factors are all re-scaled to 6% annualized volatility, except for equity which is re-scaled back to its original volatility. At a 6% annualized volatility, the Fixed – Duration factor approximately 1/10th the duration in years. For example, a 0.2 beta to the Fixed – Duration factor is roughly equivalent to a duration of 2 years. Volatility scaling is performed both when creating the risk model (at the factor level) as well as when performing the factor analysis (at the asset level). After the analysis, the asset is re-scaled back to its original volatility. This normalization helps to ensure parameters like volatility do not corrupt the data. The factors are all re-scaled to 6% annualized volatility, except for equity which is re-scaled back to its original volatility. At a 6% annualized volatility, the Fixed – Duration factor approximately 1/10th the duration in years. For example, a 0.2 beta to the Fixed – Duration factor is roughly equivalent to a duration of 2 years.

### Factors in the Global Multi-Asset Risk Model:

`In[2]: with riskmodel.global as rm:`

for f in rm.factors:

print(f.name, f.description)

**Alt – Carry**

The return generated from holding high-interest currencies against low-interest currencies

**Alt – Commodities**

Spot price returns of various baskets of commodities

**Alt – Dollar**

The US Dollar against a basket of developed market currencies

**Alt – HF Crowding**

Concentration risk from 13-F filings made by top hedge fund managers. Begins in October, 2004

**Alt – Illiquidity**

Illiquidity risk inside regularly traded names based on market dynamics

**Alt – Trend**

Risk related to trend-following factors across macro markets such as equity indices, interest rates, commodities and currencies

**Eq – Defensive**

Quality and low volatility factors, additional return by owning companies that are “safer”

**Eq – Momentum**

Risk from owning high momentum equities over low momentum equities

**Eq – Real Estate**

Additional risk from real estate securities

**Eq – Small Cap**

The additional return generated from owning smaller companies

**Eq – Value**

The value premium associated with owning high value (e.g. low price/book ratio) stocks over low value (e.g. high price/book ratio) stocks

**Equity**

Market capitalization weighted equity returns in local developed and emerging markets across the globe

**FI/Eq – Emerging Markets**

Emerging Markets exposure

**Fixed – Credit**

Credit risk in the corporate bond space

**Fixed – Duration**

Interest rate risk from changes in the yield curve. A multiple of 10 to this factor is a great approximation for interest rate duration (in years).