Optimizing Optimization: The Next Generation of Optimization Applications and Theory (Quantitative Finance)

(Romina) #1

112 Optimizing Optimization


optimization. It is especially suitable for assets with nonnormal return distri-
butions, which is common for assets such as hedge funds, private equity, and
commodities.


Portfolio attributes

Efficient surface

The WPA shows a three-dimensional plot of the current portfolio, the bench-
mark portfolio, and 10 optimal portfolios that lie on the efficient surface. The
vertical axis is expected return, and the other two dimensions are expected risk
(standard deviation) and tracking error.
The user can identify portfolios that have:



  1. The same expected return with lower standard deviation

  2. The same risk (standard deviation) with higher expected return


Higher moments

The WPA shows a variety of portfolio attributes for the various portfolios,
including skewness, kurtosis, and probability of breaching a user-specified thresh-
old. These results are particularly useful for comparing an existing portfolio to a
full-scale optimal portfolio.


Probability of loss

The WPA allows the user to evaluate probability of loss in a variety of ways.
It shows probability of loss both at the end of the horizon and throughout the
horizon. The user has the flexibility to vary the length of the horizon, the per-
centage loss threshold, and the portfolio value.
The WPA allows the user to estimate probability of loss analytically, by
Monte Carlo simulation or by bootstrap simulation.


Value at risk

The WPA shows value at risk measured both at the end of the horizon and
throughout the horizon. The user has the flexibility to vary the length of the
horizon, the VaR threshold, and the portfolio value.
The WPA allows the user to estimate value at risk analytically, by Monte
Carlo simulation or by bootstrap simulation.


Joint probability of loss

Many investors care about both absolute and relative performance. The WPA
shows the joint probability of failing to achieve an absolute target and simulta-
neously underperforming the benchmark.

Free download pdf