Statistical Process Control in Financial Markets
Tutor / director / evaluatorVan Vliet, Benjamin
Document typeMaster thesis (pre-Bologna period)
Rights accessRestricted access - confidentiality agreement
This Master’s Thesis consists of the definition, identification and calculation of several tools of Statistical Process Control applied to trading/investment systems. As Benjamin Van Vliet described in his book Quality Money Management: “a trading/investment system consists of the interacting position selection and execution algorithms, that is, the rules and business logic necessary to enter into and exit from positions in the financial markets, as well as the technology required to partially or fully automate the trading, benchmarking, portfolio, and risk management processes”.  A trading/investment system can never achieve an in-controlled status because of the uncontrollable variations of the inputs from the real word. The instability of inputs leads to variation of the outputs factors of the system just like the manufacturing processes. Good trading/investment system design needs research into the past market movements in order to analyze, optimize and validate a trading/investment system (backtesting). A good optimization and backtesting of the system will not only confirm the validity and accuracy of a system’s algorithm, but also its performance and process variation. The study considers the weekly return of several stocks from the New York Stock Exchange (NYSE): IBM, Microsoft, General Electric, General Motors, McDonald’s, Coca-Cola Company, Wal-Mart, Boeing and 3M. For this study we have worked with historical data from the 01/10/2005 to the 11/13/2008, a total of 970 measures of closing prices obtained in Yahoo Finance. We have run SPC tests with MINITAB to see if the weekly returns were out of control. As it is well-known, good inputs are the key to success in financial modeling, and forecasting, as well as risk management, requires good, clean data for successful testing and simulation of the trading/investment systems. We have analyzed distributions graphically with plots and histograms in order to determine the quality of the data. We have also performed statistical tests of preprocessing, and described cleaning algorithms to emphasize how important they are in financial modeling.