Inside the Black Box: The Simple Truth About Quantitative Trading

4.2. LIMITING THE TYPES OF RISK

Though limiting the amount of an exposure is important, some approaches to risk modeling focus on eliminating whole types of exposure entirely. Imagine that an investor's analysis indicates that CVX is likely to outperform XOM. But the trade the investor makes is simply to go long CVX while ignoring XOM. If the market drops precipitously afterward, it is likely that the investor will lose money on the trade, even if his original thesis proves correct. This is because the investor is exposed to market directional risk, even though he didn't have any particular foresight as to where the market was going. The investor could have substantially eliminated the unintentional or accidental market direction risk if he had expressed his analysis by buying CVX and shorting an equivalent amount of XOM. This way, whether the market rises, falls, or does nothing, he is indifferent. He is only affected by being right or wrong that CVX would outperform XOM.

As a general rule, it is always better to eliminate any unintentional exposures, since there should be no expectation of being compensated sufficiently for accepting them. Quantitative risk models designed to eliminate undesired exposures come in two familiar flavors: theoretical and empirical. Each is discussed in detail later in this chapter.

It is also worth noting that alpha models can (and often do) incorporate risk management concepts. Let's assume that a quant is building a relative alpha strategy. A ...