配對交易策略應用於台灣股票市場之實證研究

Pairs trading strategy applied in Taiwan stock market

林奇生(Chi-Sheng Lin)

指導教授 ： 謝文良

淡江大學/商學院/財務金融學系碩士在職專班/碩士(2006年)

https://doi.org/10.6846/TKU.2006.00872

摘要

在國外，配對交易屬於統計套利策略中最普遍為市場所運用者，本研究之目的在於運用計量模型為分析工具，並以台灣股票集中市場之台灣50成分股為研究對象，望能提供投資人有關配對交易策略能否有效運用於台灣股票市場之有用參考資訊。
本研究在挑選配對股票之方法係先對所有研究變數進行關聯分析，以兩研究變數之相關係數高於+0.6或低於-0.6者為第一道篩選關卡，符合條件者即列為可進行配對交易之觀察名單。凡出現交易訊號者，再以定態及共整合為基礎，確立此一配對股票確實存在共整合關係，具有回復至長期平均值之特性。估計期與交易期之取樣期間為2003年5月1日至2005年4月30日止，共涵蓋兩年。
經由實證結果顯示，在相同交易策略下，類股組合之年化報酬率分別為25.4%以及114.4%，遠優於產業組合之8.9%年化報酬率。究其原因，應為本研究於進行配對交易時加入了共整合檢定以及基本面條件，即使非屬同一產業，當股價彼此偏離過大時，將誘使套利買盤進場，讓股價又重回穩定的均衡關係。

關鍵字

對沖基金 ； 避險基金 ； 配對交易 ； 統計套利 ； 回復均值

並列摘要

In other countries, pairs trading is the most popular strategy in statistical arbitrage.The purpose of this study is using pairs trading strategy to analyze Taiwan 50 index constituents and seeing if this strategy is a proper and effective way for the average investors who like to gain a new perspective in Taiwan stock market.
The most expedient way for pairs trading is correlation analysis. The list of pairs is explicitly partitioned into two sets: potentially cointegrated and not potentially cointegrated. We could just as well work with a candidate list of all possible pairs, run cointegation tests on all of them, and eliminate pairs that fail the tests. The formation period and trading period is between May first 2003 and April 30th 2005.
The result appears that annualized return of sector portfolio is 25.4% and 114.4% in different trading strategy. It is apparently prior than the of sub-sector portfolio, which is 8.9%. After considering the approach of using cointegration testing and fundamental analysis, even though the pairs are in the different industries, if there is a deviation from the long-run mean, one or both time series will adjust themselves to restore the long-run equilibrium.

並列關鍵字

hedge fund ； pairs trading ； statistical arbitrage ； mean reversion

參考文獻

(1)Akaike,H.(1974),A New Look at Statistical Model Indenication,IEEE Transactions on Automatic Control AC-19-6,pp716-723

(2)Brockwell,P.,and Davis,R.(1991),Introduction of Time Series and Forecasting,Section9.3,New York:Springer-Verlag

(3)Dickey,D.A.and Fuller,W.A.(1979),Distribution of the Estimators for Autoregressive Time Series with a Unit Root,Journal of the American Statistial Association, Vol.74,pp427-431

(4)Dickey,D.A.and Fuller,W.A.(1981),Likelihood Ratio Statistics for Autoregressive Time Series with a Unit Root,Econometrica,Vol.49,pp1057-1072

(10)Gonzalo,J.(1994),Five Alternative Methods of Estimating Long-Run Equilibrium Relationships,journal of Econometrics,Vol.60,pp203-233