Dynamic Physical Behavior Analysis for Financial Trading Decision Support [Application Notes]

This article aims to discuss the application of computational intelligence (CI) techniques in combination "with classical concepts in physics in devising investment strategies. In the analysis of investment strategies, many CI techniques are employed to predict market trends, such as the neural network (NN), the support vector machine (SVM), and particle swarm optimization (PSO) techniques. Other techniques such as evolutionary computing (EC) and genetic algorithm (GA) are utilized to identify the knowledge rules of trading. However, changes in market behavior are dynamic and time variant. Thus, using a single CI technique can occasionally be better than traditional statistic models, but the trading models may pose risks from the changing market. Recently, the hybrid model and the data mining concept, "which combine multiple CI techniques into multiple stages, have emerged to improve the trading model\´s stability and profitability. For example, fuzzy logic is employed to differentiate the parameters in the first stage, and then similarity search is used for data clustering in the second stage.