Final Submission, 2022.pdf (7.29 MB)
Download file

Forecasting stock prices on the basis of technical analysis in the industrial sectors of the UK stock market

Download (7.29 MB)
thesis
posted on 15.03.2022, 17:15 by Md Aminur Rahman
This study aims to evaluate, critically and rigorously the weak-form market efficiency and forecasting power of technical analysis in different industries in the developed market, the London Stock Exchange. For this purpose, weekly data were collected from the FTSE-all share index, FTSE-350 general industrial index and twenty companies of four different industries, for the period between 3 March 1997 and 16 July 2017. Friedman test detects that data do not have any seasonality. It is found that there is a statistically significant impact of structural breakpoints for all selected series. Thus, their impacts have been ignored for the purpose of analysing and forecasting stock prices through identifying the plain data of sub-sample periods where there is no structural break for each series from the application of Bai-Perron’s multiple breaks test. The descriptive statistics table, histograms, and kernel density graphs show that the weekly closing prices of all the selected series are not normally distributed. The runs test and variance ratio tests document that the stock prices of all the series do not change erratically and randomly. Furthermore, Ljung-Box’s serial autocorrelation test evidences that a few series do not have serial auto-correlation at the first difference. Moreover, the ADF-unit root test demonstrates that there is no unit root at the first difference for all series. Therefore, the statistical inference was made that the market is not weak-form efficient in the period of the tests for all series and their stock prices are predictable. This study extends the current literature by considering the existence of weak-form inefficiency in different industrial sectors. It is found that industrial sectors impact market efficiency. The ARIMA and GARCH (1, 1) models contribute to evidence better short-term predictability of stock prices for all series in the manufacturing industry. Contrary to that, these models argue that most of the series in the service industry are not predictable. Therefore, this study does not find any support for weak-form efficiency over the periods tested in the London Stock Exchange. The ARIMA model shows superior prediction power to the GARCH (1,1) model and exponential smoothing techniques for most of the series. On the other hand, exponential smoothing techniques perform even better than ARIMA and GARCH (1, 1) models for a few series in the manufacturing industry due to series or industry characteristics. It is found that certain econometric models are better in certain industry sectors

Funding

Self-funded

History

Year

2022

School

School of Management