Vinga Asset Allocation
Vinga Asset Allocation is an add-on to EcoWin Pro
and is a fundamental enhancement of EcoWin Asset Allocation launched in
the year 2000.
The fundamental core of the Mean Variance Approach is that financial instrument returns can be measured in two dimensions, expected returns and level of return uncertainty in the return i.e. returns and risks.
The expected returns can be measured as the expected value of the uncertain returns. The degree of uncertainty, the risk, is measured as the standard deviation of returns, or the square of the standard deviation, the variance.
The investor is supposed to be risk-averse, i.e. that a higher risk must give a higher expected return. In the choice between two investments with the same expected returns the one with the lower risk is preferred. When the choice is between two alternatives at same risk level, the one with highest expected return is preferred.
The next natural step is, from the set of possible portfolios, to point out the portfolios that risk-averse investors find better then others. This means to describe the set of efficient portfolios. The set can be defined with or without restrictions on the proportions of the included instruments.
The difficulty with an operational application of the above analysis is to find information on the expected return and the risk of instruments.
The most common way to forecast these quantities is to use historical information on instrument returns. The fundamental assumption behind using such estimates is that these are applicable also for the future, i.e. over the horizon on which the forecast is done. In Vinga Asset Allocation both expected returns as well as risks are can be estimated from historical information. Covariance matrices and expected return vectors can be pasted into the application and the optimization can be based on these.
The application now launched contains a number of novelties:
The uncertainty of estimated risks and expected returns is faced by solving the optimization problem for a number of simulated covariance matrices and expected return vectors. The simulation is made by creating random scenarios out of the historically observed instrument returns or from normally distributed instrument returns with the original distribution found in historical data.
Covariance matrices can be imported to the application and from all matrices noise-reduced matrices can be constructed using the principal component method.
As an integrated part also Value at Risk and Expected Shortfall can be calculated for user specified portfolios. The Value at risk is the loss such that for some assigned probability losses are smaller. The expected shortfall is the expected loss given that losses are larger than VaR. Other names for ES are Conditional value at risk or Tail loss expectation.
The Value at Risk has become an industry dominant indicator of risk. However, recent research has revealed some fundamental drawbacks of this measure. Some set of reasonable properties of any risk measure is defined in the concept of Coherency. The VaR is not a coherent measure, while Expected Shortfall is.
VaR and ES are based on either the historical empirical distribution of losses or on covariance matrices and expected values of losses. Further, the uncertainties of these measures are presented as distributions generated by repeated simulations.
Below are some screenshots taken from the Vinga Asset Allocation Addon.