中文摘要：

基于转移模型，收集中国A股市场开展融资融券业务的900只股票和15只ETF的数据，采用logt检验和俱乐部收敛检验研究A股市场属于同一行业的融资融券标的股票价格以及融资融券标的ETF价格之间的收敛关系及其动态变化，分析A股市场统计套利的风险，研究发现基于股票所属行业构建套利组合具有较大风险，融资融券的双向交易机制尚未充分发挥，市场有效程度不足，但统计套利在中国市场仍具有可行性。实证结果表明，从整个行业看，融资融券标的股票之间不存在持续、稳定的收敛关系，但同一行业内某些个股之间存在不稳定的俱乐部收敛关系；融资融券标的分阶段扩容增加统计套利的风险，也增加统计套利的机会。扩容之后绝大部分行业个股整体收敛关系消失，但俱乐部数量增加，俱乐部收敛检验可以为选择套利资产组合提供依据；考虑新增标的在内，ETF整体在样本期间内具有稳定的收敛关系，更适合作为套利资产组合。样本外检验证实了前述结果。研究结论为对冲基金进行统计套利风险分析提供经验证据，并为选择套利资产组合提供方法。

英文摘要：

Statistical arbitrage, theoretically is riskless, however the prices of the assets might diverge, forcing investors who exploit the relative mispricing of assets to unwind some of their positions at spectacular loss. Convergence of the arbitrage assets＇ prices is a fundamental issue of statistical arbitrage since the unsteady convergence relationship between arbitrage assets causes arbitrage risk, cost as well as expected return of arbitrage. Based on transition model proposed by Phillips and Sul （2007） , this paper studies convergence among the securities eligible for margin trading in A share market, and analyzes the risk of statistical arbitrage. With the empirical study on the trading data of all the securities eligible for margin trading from April 2010 to June 2015, we find constructing portfolio solely by stocks＇ sector is risky, and high risk of arbitrage implies the market lacks efficiency as the market is inefficient in correcting the misprieing, which is consistent with intuition. However, statistical arbitrage in A-share market is feasible. Result shows that the stocks of identical sectors as a whole do not converge steadily or robustly. Stocks of manufacture, real estate, non-bank finance, etc. converge as a whole through the first phase, but after the expansion of the eligible list of mar- gin trading, convergence in the whole sector diminishes. However, this does not deny the feasibility of statistical arbitrage, the phase-in expansion of eligible list brings opportunities as well as risks since more eligible stocks increase both diversity and complexity of asset selection as some newly added stocks converge with the existing stocks while others do not. By testing the convergence within the sectors, we find certain stocks converge and form subgroups which are referred as club convergence in this paper. Clubs increase with the expansion of eligible list although they do not increase strictly across different phases. Compared with stocks, the ETFs as a whole group converge robustly and the ne