Multifactor models for Equity and F.I.
We have built from scratch both equity and fixed income factor models for the Indian stock and bond markets.
In the equity risk models, stock returns are dependent on the exposure of the stocks to risk indices and industry groups. Examples of these risk indices are value, growth , momentum and variability. These risk indices themselves are made up of several basic building blocks. For example, the variability factor is composed of the stock alphas, their betas and their sigma's, say. These are combined together to create the overall risk index. The risk indices are themselves standardized so the market as a whole will have zero exposure to the risk index and the variance of the exposure is unity. Stock returns are therefore composed of the returns due to these risk indices and the industry groupings. The risk associated with each stock is therefore due to the risk factors and the industry groups as well as an idiosyncratic part.
The risk model has several useful roles. First, we can easily decompose the risk associated with the returns of any stock or a portfolio of stocks into the risk due to the risk indices and industries and the specific risk. This decomposition is very useful in assessing the risks associated with a particular stock portfolio and in controlling unwanted risks. Second, the risk model provides the input for an optimizer where many different types of portfolios can be constructed while taking into consideration the risk-return tradeoff. For example we could try to replicate a known index using a smaller number of assets while controlling for the risk of the portfolio. Or we could set up sector rotation strategies using the optimizer. Third, we can try to forecast the returns to the common factors themselves and use these forecasts to create an active portfolio that will outperform a benchmark. This forms the basis for what are known as style strategies.
The fixed income model also relies on a factor based approach. The focus here is, of course, interest rates. We model not just the level of rates but also the effect of twists and non-parallel moves in the temstructure. In addition we model the spreads associated with lower quality instruments . We also look at the effect of built in option features and F/X risk associated with bonds, if these are applicable.
The focus in all this modeling is our disciplined quantitative approach. We employ state of the art modeling techniques in creating these models.