Description

COMMODITIES PORTFOLIO OPTIMIZATION

The example presents an investor’s decision. Investor wants to invest $200,000 on a portfolio of four different metal commodities:  gold, silver, platinum and palladium. His portfolio horizon is two years. He is not decided on which type of risk profile he should lean to. Our objective is to present him with alternative portfolios that would optimize his choice on how to invest upon the four different commodities presented here.

The objective of this module is to show an example containing two powerful features of @RISK:  time series tool and RiskOptimizer, its optimization tool.  We will use time series analysis to forecast the future stochastic price of four different metal commodities 24 months upfront considering historic price variations and the correlations between these assets.

Then, we will use optimization techniques using RiskOptimizer in order to determine different portfolios, that is, combinations of assets that will comply with the investor’s risk profile objectives.

TIME SERIES

In the fields of statistics, economics, and mathematical finance, a time series is a sequence of observations, typically measured at regularly spaced times, such as every week, every month, or every quarter. We will use here an example of commodity prices for metals. Other examples of time series are weekly currency exchange rates, the daily closing value of the NASDAQ Composite index, and monthly crude oil prices or any other type of commodity.

RISKOPTIMIZER

@RISK includes RISKOptimizer, a powerful tool that combines simulation and optimization to find optimal solutions to models that contain uncertainty. This is the example we are presenting here. The uncertainty is present in this model by the time series segment that considers all possible values for the portfolio. By using optimization techniques and Monte Carlo simulation, RISKOptimizer can find optimal solutions to problems that cannot be solved by standard linear and nonlinear optimizers.

In that sense, RISKOptimizer combines the simulation technology of @RISK and the optimization engines of Evolver. This latter software employs two major algorithms: Palisade’s own genetic algorithms and OptQuest, a widely used optimizer, to solve deterministic optimization models.  RISKOptimizer takes the process one significant step farther by optimizing models with explicit uncertainty such as this example.