配對交易是一種利用兩支股票之間均衡定價的短暫隨機偏離進行操作的統計套利方法。本研究分為兩個部分，並使用S&P 500的股票進行實證分析。第一部分運用日資料於共整合模型，並且預先指定臨界值以建構配對交易，後續我們運用主成分分析、切片逆迴歸法及核化切片逆回歸法調查特定的股票變數(如每股盈餘、共整合強度等)對於配對交易利潤的影響，研究結果發現由選定的股票變數中能挖掘有風險的區域，投資者選取風險區域中的配對無法有效地進行獲利。
第二部分使用日內資料建構高頻配對交易，同時我們選取數個高頻共變量(如流量指標、相對強弱指數等)作為支持向量機分類器的輸入特徵，經由變數訓練的分類器建立進場的交易訊號，我們將探討此交易訊號下的配對交易之表現。

Pairs trading is a comparative-value form of statistical arbitrage designed to use temporary random departures from equilibrium pricing between two stocks. In the first part, we use the spreads of cointegrated pairs and pre-chosen thresholds to perform pair trading for daily data. We investigate the effects of several selected covariates (e.g. EPS, strength of cointegration and etc.) on the pairs trading profits. We use principal component analysis, sliced inverse regression and kernel sliced inverse regression to find risky covariate zones which result in unprofitable pairs. In the second part, we conduct high-frequency pairs trading for intraday data. We use several high frequency covariates (e.g. money flow, relative strength index and etc.) as input features for support vector machine classification to set up trading signals of entering positions. We investigate the performance of the proposed pairs trading strategies for stocks in S&P 500 index.

誌謝 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .i
摘要 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . ..ii
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . iii
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .1
2 Pairs Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2
2.1 Mean Reversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . 2
2.2 Cointegration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . 2
2.3 Johansen Test for Cointegration . . . . . . . . . . . . . . . . . . . . . . .. . . 3
2.4 Standardized Spreads . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . 4
3 Data Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . 5
3.1 Data Acquisition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . 5
3.2 Covariates for Daily Data . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . 6
3.3 Covariates for Intraday Data . . . . . . . . . . . . . . . . . . . . . . . . . .. . . 8
4 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .9
4.1 Principal Component Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . .9
4.2 Sliced Inverse Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . 11
4.3 Support Vector Machine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
5 Trading Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .18
5.1 Trading Strategy for Daily Data . . . . . . . . . . . . . . . . . . . . . . . .. . . 18
5.2 Trading Strategy for Intraday Data . . . . . . . . . . . . . . . . . .. . . . . . 19
6 Empirical Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .20
6.1 Performance for Daily Trading . . . . . . . . . . . . . . . . . . . . . . . . . . . .20
6.2 Analysis of Pair Trading Result . . . . . . . . . . . . . . . . . . . . . . . .. . . 22
6.3 Performance for Intraday Trading . . . . . . . . . . . . . . . .. . . . . . . . . 23
7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . 24

Chen, N. H. (2010). emph{Time Series: Applications to Finance with R and S-Plus, 2nd ed.} John Wiley & Sons, New Jersey.
Chiou, M. C. (2015). emph{Analysis of Variables Affecting Pairs Trading Profits.} Master thesis, Department of Applied Mathematics, National Sun Yat-sen University, Kaohsiung.
Cortes, C. and Vapnik, V. (1995). Support-vector networks. emph{Machine Learning}, extbf{20}, 273-297.
Duan, N. and Li, K.-C. (1991). Slicing Regression: A Link-free Regression Method. emph{Annals of Statistics}, extbf{19}(2), 505-530.
Engle, R. and Granger, C. (1987). Cointegration and Error Correction: Representation, Estimation, and Testing. emph{Econometrica}, extbf{55}(2), 251-276.
James, G., Witten, D., Hastie, T. and Tibshirani R. (2013). emph{An Introduction to Statistical Learning - With Applications in R}. Springer, New York.
Johansen, S. and Juselius, K. (1990). Maximum Likelihood Estimation and Inference on Cointegration - With Applications to the Demand for Money. emph{Oxford Bulletin of Economics and Statistics}, extbf{52}(2), 169-210.
Johnson, R. A. and Wichern, D. W (2007). emph{Applied Multivariate Statistical Analysis, 6th ed.} Pearson, Essex.
Li, K. C. (1991). Sliced Inverse Regression for Dimension Reduction. emph{American Statistical Association}, extbf{86}(414), 316-327.
Wu, H. M. (2008). Kernel Sliced Inverse Regression with Applications to Classification. emph{Journal of Computational and graphical statistics}, extbf{17}(3), 590-610.
Wu, P. S. (2015). emph{Trading Strategy Based on Cointegration Pairs.} Master thesis, Department of Applied Mathematics, National Sun Yat-sen University, Kaohsiung.