The Evaluation and Optimization of Trading Strategies, 2nd Edition

CHAPTER 9

Search and Judgment

The next stage in the evaluation of a trading strategy is optimization, which is presented in the next chapter. Before we can proceed with a presentation of these methods, however, we need to explore a little understood but extremely important aspect of this process: the search of the optimization space.

To fully understand the most effective way optimization can be applied to a particular trading strategy, we need to understand the contours of this seemingly innocuous area. And before we can fully understand how we can arrive at those trading strategy parameters that are most likely going to produce real-time and lasting trading profits, we need to understand the all-important impact of what has typically been known as the optimization criterion. In optimization theory this is also known as the fitness function, objective function, search criteria, and search function. In conformity with contemporary usage, we will use the term objective function as synonyms for all of these other terms.

To elaborate, the optimization space is simply the set of all possible strategy parameters, or candidates, for which a historical simulation can be created. An optimization must be performed on multiple pairs of different moving average lengths to determine the most profitable combination of moving averages for our moving average example. To do this, for example, an optimization is set to scan, or create, a simulation for every combination of moving average lengths ...