Based On Neural Network Technology To Analyze The Effectiveness Of Research

In these decades technical analysis as one of the most important parts of financialanalysis, gets more and more piers with the development of information technology.Since Chinese stock market opened interests and researches on technical analysis growrapidly and investors pay attention on this area more wildly.Firstly, we makes a systematic review of core researches about market efficiency,technical analysis efficiency and related concepts. After a wide theoretic research and anempirical research on Chinese stock market, we get the conclusion that research oftechnical analysis in Chinese stock market is necessary and uesful.Based on this conclusion, we proposed a systematic and automatic model to predicttechnical indicators using neural networks which was optimized by BP algorithm. In thismodel, back propagation algorithm based on forward networks was conducted to learninformation of historical data and to train the network weights.Secondly, We apply a single-hidden-layer back-propagation neural network to theindex in Shanghai Stock Exchange of China. By using Momentum Strategy as tradingrule, the results show than the neural network model can get better returns and lessvolatile than the random walk model. At the same time, we can find an abnormal profitcan be achieved. After the use of bootstrap techniques as statistical analysis the abnormalprofit is also significant. We also find a good forecasting performance compared withsimple momentum strategy after we calculate NMSE, Grad and Sign which are wildlyused as the criterion for valuation.Thirdly, we also used a single-hidden-layer back-propagation neural network tocapture the relationship between the contrarian strategy and the sample stocks of the 180index in Shanghai Stock Exchange. The results show that an abnormal profit can beachieved. After the use of bootstrap techniques abnormal profit forecasted by the neuralnetwork mode is also significant, this confirms the contrarian strategy based on neuralnetwork model is efficient for forecasting in stock market in China. Moreover, we findthat the constrarian strategy performs best in Fluctuating market but worst in Bearishmarket, which shows investors maybe need different strategy in different marketcondition.