Abstract:

We cannot reliably predict how the price of a single asset will fluctuate over time; but sometimes there are predictable patterns in the relative movements of related asset prices. The tendency of relative asset prices to move in predictable ways is called correlation. While we do not have a general theory of correlation, there are some basic examples that are important and easy to understand. In this presentation I will present some of these basic examples as motivation for some of the general modeling and analysis problems associated with correlation.

The simplest example of a correlation position is a spread trade between two assets. Basis trading, swap trading and relative-value trading are strategies for reducing risk (hedging) when the spread is fairly priced, or for making money (speculating) when the price spread between pairs of assets move out of line.

In the standard course of business, financial institutions often find it advantageous to trade some of the assets they naturally accumulate for other types of assets in order to diversify their risk. Correlation models are necessary to quantify the benefits of these hedging and risk management activities.

When dealers find themselves stuck with large levels of correlated portfolio risk that they cannot easily hedge (e.g. catastrophic risk), they sometimes repackage and offload some of this risk in the form of a derivative product. The counterparty (typically a hedge fund) will want an attractive expected payoff for taking the other side of the trade. One example of such a product is a correlation swap. On account of financial engineering inventions like these, correlation trading has become a growing market with its own set of specialized, evolving products.