Ralf Scholl

Scientific Computation: Stock market prediction

Topics Dynamic programming Optimization of decision-based functionals Simulation: modelling, validation, prediction Introduction BASIC Stock market prediction is a very interesting topic and could also be very profitable if done successfully. It falls well within our definition of modelling/prediction. We define “stock market prediction” as the techniques which extract information from past behavior to decide when to buy or sell which stock and at what price. This is called quantitative analysis or technical analysis, i.e. the use of mathematics to predict future behavior based on historical data. It is normally contrasted with fundamental analysis which analyses the health of the business in a very broad sense to predict the future behavior of stock prices. There are several reasons which make technical analysis a very good topic for the study of methods in modelling and prediction. (i) The problem is simple and easy to define precisely. (ii) All the main concepts of modelling/prediction appear in this problem. (iii) There are abundant data – most publicly available through the World Wide Web. (iv) New data are generated every day, which allows extensive and realistic validation. Please note that the goal of this chapter is to illustrate methods in modelling and simulation. It is not a complete or up-to-date treatise on stock market prediction. Definitions We are going to use only buying/selling of regular stock and avoid more sophisticated tools such as options, futures etc. A few definitions are necessary to clarify the concepts.