本研究主要目的為預測共整合配對交易之獲利。共整合配對交易為統計套利的交易策略，其獲利性來自共整合誤差具有均數回歸的性質。為達預測之目的，本文利用蒙地卡羅模擬法，重複生成具共整合的股票價格以執行模擬交易，將多次模擬獲利之平均做為預測値。在本文中分別根據以殘差為基礎之共整合法與向量誤差修正模型兩種方法，對歷史資料來建立具有共整合關係股票價格的模型，並藉此模型生成未來的股票以用於模擬交易。實證資料使用納斯達克100 指數成分股做為資料集，使用上述兩種方法對所有組合股票價格辨識是否配對具共整合關係與建立模型，並比較蒙地卡羅模擬法之模擬結果與實際交易結果。結果顯示，以殘差為基礎之共整合法的模擬結果無法有效解釋實證結果。而在向量誤差修正模型實證獲利結果分為正獲利群與負獲利群，兩群與各自對應之模擬獲利皆呈線性趨勢。因此，將兩者各自配適迴歸線，且兩模型解釋能力約達78% 以上。平均來說，正獲利方面，模擬獲利低估實證獲利，負獲利方面，模擬獲利取負號為高估實證損失。此外，我們以模擬之平均交易所需時間做為配對篩選，使挑選配對之獲利總和為正。

The aim of this study is to predict cointegration pairs trading profits. Cointegration pairs trading is a statistical arbitrage strategy. And the profits are due to the mean reverting property of cointegration error. In this study, we adopt Monte Carlo method to predict the profits. We generate the prices which are cointegrated and perform the pairs trading to obtain the simulated profit. By repeating the above procedure, the average simulated profits are regarded as the predicted profits.The vector error correction model and residuals-based cointegration approach are used to identify cointegration relationship and generate the cointegrated data.In empirical study, NASDAQ 100 index components are chosen as database. We perform cointegration pairs trading by above two approaches and then compare the real and simulated profits. The result shows that the residuals-based cointegration approach cannot reveal the real situation. Alternative, the vector error correction model can divide real profits into two groups: the positive and negative profits group.In order to describe the linear relationship between simulated profits and real profits, each group is fitted by a linear regression line with corresponding simulated profits respectively and $R^2$ of fitted model are about 78%.Furthermore, the average of trading duration time is used to select trading pairs. It allows us to filter the negative profits.

1. Biondini, R., Lin, Y. X., and McCrae, M. (2003). A case study of the residual-based cointegration procedure. Journal of Applied Mathematics and Decision Sciences, 7(1), 29-48.
2. Chan, N. H. (2011). Time series: applications to finance with R and S-Plus, 2nd edition. New York: Wiley.
3. Engle, R. F., and Granger, C. W. (1987). Co-integration and error correction: representation, estimation, and testing. Econometrica, 55(2), 251-276.
4. Figuerola-Ferretti, I., Paraskevopoulos, I., and Tang, T. (2014). Pairs Trading and Relative Liquidity in the European Stock Market. Unpublished manuscript.
5. Granger, C. W. (1981). Some properties of time series data and their use in econometric model specification. Journal of econometrics, 16(1), 121-130.
6. Johansen, S. (1988). Statistical analysis of cointegration vectors. Journal of Economic Dynamics and Control, 12(2), 231-254.
7. Johansen, S. (1991). Estimation and Hypothesis Testing of Cointegration Vectors in Gaussian Vector Autoregressive Models. Econometrica, 59(6), 1551-1580.
8. Lin, Y. X., and McCrae, M. (1999). A new approach of determining cointegrating vectors. MODSIM99 Proceedings, University of Waikato, Hamilton, New Zealand, 265-270.
9. Maddala, G. S., and Kim, I. M. (1998). Unit roots, cointegration, and structural
change(No. 4). Cambridge University Press.
10. Rebonato, R., and Jackel, P. (1999). The most general methodology to create a valid correlation matrix for risk management and pricing purposes. Journal of Risk, 2(2), 1-12.
11. Vidyamurthy, G. (2004). Pairs Trading: quantitative methods and analysis (Vol. 217). New York: Wiley.