Statistical Analysis Of Stock Technical Indicators

Technical analysis is an investment method to forecast the stock prices trends by the study of past information of the stock, which originated from the Dow's theory. Now it has formed the two major investment styles together with the fundamental analysis after more than a hundred years' practice, and technical indicators are its indispensable tools. Though it's very rare for investors to give up technical indicators in practice completely, technical analysis has given people an empirical impression for a long time because of its lack of scientific theory, therefore the discussion on the predictability of is technical indicators is always a hotspot. The early research indicated that technical analysis is actually useless, but the recent studies suggest this conclusion might have been premature. This thesis investigates two popular technicalindicators——RSI and BIAS by theory and positive analysis respectively, to exploretheir statistical properties and to verify weather they could make a successful forecast.Theoretically, on the hypothesis of that stock prices follow geometric Brownian motion, we get that the series of RSI and BIAS are strictly stationary and m-dependent, and their partial sums are asymptotic. The ideal hypothesis implies the stock prices satisfy Markov property which is consistent with the EMH that means technical analysis is in vain. There is rare literature to investigate the statistical properties of technical indicators in theory.In the positive analysis, by utilizing the Standard Pool Index and Hang Seng Index trading data, various trading rules of two indicators are tested. Most of the available literature assumes the stock returns are independent and normal, which is turned out to be untrue apparently in real world. There are no such assumptions and model specifications in the positive analysis of this paper, which compares the conditional returns selected out by technical indicators to the unconditional returns by random sampling inspired by the thought of Bootstrap methodology,instead and finally concludes that as long as one chooses the appropriate combinations of parameters of technical indicators, one can gain excessive returns, and the effectivity of technical indicators is contrary to that of the securities market. The conclusions of this thesis could provide some references for researchers and investors.