| Abstract [eng] |
118 pages, 2 tables, 8 figures, 70 references This thesis focuses on empirical analysis on whether technical analysis can in fact accurately forecast the market movements of financial markets of the globe by use of arch, garch, Johansen co integration tests as well as auto regressive integrated moving average. The first purpose is to investigate the applicability of these models for measuring market risks, the extent of market integration, and their forecasting ability in developed and emerging markets. The results of the GARCH models for the ALL SHARE FTSE, Sensex, and Nifty 50 indices describe the presence of clusters of high volatility. Volatility clustering means that high or low volatility will continue for an extended period, thus, these indices are perfect for technical analysis. This finding is consistent with the basic assumptions of technical analysis that entails the belief that the price movements of securities in the future will mirror those of the past. On the other hand, it may be illustrated that the indices such as UK FTSE 100 exhibits low GARCH coefficients and this imply that the volatility has low sample auto correlation hence it does not have strong persistence. This brings the difficulty in trying to forecast market behaviour on such indices, thereby underlining the weakness of technical analysis in the market with low volatility persistence. In the long-run aspect, Johansen cointegration test is used for determining the connection of many global financial indices. The findings show highly significant evidence of cointegration in the developed markets like Sensex and ALL SHARE FTSE which imply that the prices of the two markets are interrelated with each other. This finding also suggests that these interdependent relationships may lead to the emergence of cross-market arbitrage. Nevertheless, there is low cointegration in trading volume, implying that whereas technical analysis based information flows is almost universally valid, trading volumes have stronger components that depend on specific country effects thus limiting the universal usage of trading volume data in technical analysis. The time series forecasting models that include ARIMA stress autoregressive patterns in developed markets. In indices such as the Sensex and ALL SHARE FTSE, the past behavior of the markets actually gives the future behavior of the prices. However, moderate auto-regressive features are exhibited by emerging market SES and NIFTY fifty making short-term forecasting even more difficult due to more noises in the markets and less pattern forecastability. The study finds that technical analysis works well exceedingly in more active, interrelated markets with clustered volatility. Nevertheless, it is not as efficient in low volatility or high efficiency areas where trends can be difficult to recognize. As the study is places emphasis on the variations of technical analysis applicable to various markets, it argues that the maturity and persistence of the volatility as fundamental criteria in understanding the effectiveness of technical analysis strategies. |
