A Research Into Moving Average Price Series Pairs Trading Strategy Based On Cointegration

Moving Average formulated by Joseph Granville has been one of the most famous theories amongst numerous technical analyses in the stock market of the world. The theory mainly focuses on the relationship between closed price series and moving average price series and consists of eight principles including four buyings and four sellings. As mentioned above, we learn that the theory studies principally the connections between closed price series and moving average price series in order to search for buy point and sell point. Moreover there are kinds of moving average price series in accordance with computing cycle and they are all derivatives of closed price series. Then what is the connection between them? Can they provide trigger point or not? As we know, most financial time series are non-stationary, such as futures price series, options price series, etc. Stock price series which is most familiar to investors is non-stationary as well. To deal with the problem, Engle and Granger formulate a method called conintegration, which states that although some series are non-stationary, their linear combination can be stationary, namely there exists long equilibrium between them. Consequently, people can employ the ECM to analyse the dynamic nonequilibrium. Pairs trading strategy in connection with cointegration has become prevalent more and more since the early 1980s. This strategy was designed by a team of scientists including mathematics, computer sciences, physics, etc, who were brought together by the Wall Street quant Nunzio Tartaglia. The team aims at exploring trading platforms based on computer with the use of statistical methods, where the human subjectivity have no impact on the decision of making buy point and sell point. The concept of pairs trading is the coinstantaneous buying and selling of two relevant securities when they are mispriced. Profits are gained when the prices of the two securities converge. On the basis of the above, this paper intends to use cointegration methodology to identify trading pairs of moving average price series, and then analyzes the deviation of equilibrium between moving average price series in terms of ECM, and gains the empirical results and analyses of Granger causality furthermore, this paper employs the residual of cointegration regression equation to provide trigger point in form of moving average price series pairs trading. Subsequently this paper formulates a strategy and obtains the empirical results and analyses.The innovation of this paper is in that there are two differences between moving average price series pairs trading and traditional pairs trading, namely the difference of trading time and the difference of the number of underlying securities. To begin with, the traditional pairs trading is buying a security while selling another. There are two underlying securities consequently. The underlying security in moving average price series pairs trading, however, is only one. What is more, the moving average price series pairs trading finds the trigger point according to the deviation of equilibrium between the series, thus it causes the non-simultaneous trading time, i.e. to sell after the buying.In terms of contents, this paper is divided into four sections. In chapter one, it mainly introduces the research background and its current significance, and gives a general literature review of pairs trading from both home and abroad. In chapter two, this paper comprehensively looks back on the basic econometric theories, models, methodologies related to moving average price series pairs trading such as Unit Root Test, Cointegration, ECM, Granger causality, etc. In chapter three, this paper primarily presents data selection and data processing needed for empirical test, and design of trading strategy, and process of empirical test and analyses of related results. In accordance with the routine in Chinese securities market, this paper divides moving average price series into seven computing cycles of 3 days, 5 days, 10 days, 20 days, 30 days, 60 days, 125 days. This paper selects three listed companies of SHANDONG GOLD MINING CO.,LTD, BEIJING CENTERGATE TECHNOLOGIES(HOLDING) CO., LTD, TONGHUA GRAPE WINE CO., LTD in SHANGHAI STOCK EXCHANGE and SHENZHEN STOCK EXCHANGE as the samples, of which the first two companies are the number one in price increase and the number one in price decrease respectively, the price increase of the last one is mostly zero at the end of 2009, and then this paper divided moving average price series of 3 days, 5 days, 10 days, 20 days, 30 days, 60 days, 125 days into six trading pairs of three strategies, namely short term strategy including MA3&MA5,MA5&MA10, medium term strategy including MA10&MA20,MA20&MA30, long term strategy including MA30&MA60,MA60&MA125(MA is the abbreviation of Moving Average Price Series, MA3 is Moving Average Price Series of 3 days, & is equal to"and", the rest may be inferred). The trigger point of moving average price series pairs trading in this paper is the zero-crossing of residual from cointegration regression equation, i.e. we should buy the underlying security when the signs of residual alter from positive to negative, on the contrary, we should sell the secrity when the signs of residual alter from negative to positive. Moreover, the trading time is before the time of closing in lagged market day. Subsequently, this paper applies the strategy mentioned above to the K line of SHANDONG GOLD MINING CO., LTD to test the performance of the strategy. And then, it selects the best-performing strategy from short, medium, long term strategy to apply to K line of BEIJING CENTERGATE TECHNOLOGIES (HOLDING) CO., LTD and TONGHUA GRAPE WINE CO., LTD, after that, presents the empirical results and related analyses. The last section in this paper is the statements of conclusion and suggestions.The results in this paper lie in the following statements:(1) The moving average price series pairs trading don't need to spend much time to indentify the trading pairs while the traditional pairs trading have to. It just needs to identify the trading pairs between moving average price series.(2) We can arrive at the conclusion that the performance of long term strategy of moving average price series pairs trading including MA30&MA60 and MA60&MA125 is better than that of short term strategy including MA3&MA5 and MA5&MA10 and medium term strategy including MA10&MA20 and MA20&MA30.(3) Long term strategy can effectively avoid the risk from continuous price decrease in that its longer average-holding period and fewer transactions, the performance of it will be better if one can take the stop-loss, and it may completely take advantage of the chance from continuous price increase as the same reason, and obtain related return.(4) The performance of long term strategy applied to the SHANDONG GOLD MINING CO.,LTD which is number one in price increase, however, cannot be compared to the"ever-holding"strategy at all. The performance of long term strategy applied to the rests of the three samples is far better than that of"ever-holding"strategy. This proves that the long term strategy is a market-neutral strategy ,this characterization is consistent with that of traditional pairs trading, in other words no matter how the market fluctuates, the long term strategy can help investors to obtain steady return.