根据传统的资产定价理论，仅有系统性风险应当获得风险溢价，而证券的个体风险可以被分散，因此不应该被定价。但实证金融的结果表明，市场投资组合的收益并不足以解释单个证券的收益，调整市场风险后依然对证券价格有显著解释力的变量被称为因子，而这种现象被称为异常现象。 本文选取了41个发表在顶级期刊上的因子，采用统一的样本和构造多空策略的方法进行测试，发现仅有部分因子对未来股票收益有显著的解释力，这表明因子的存在性高度依赖样本和方法。由于过度反应的存在，因子与未来股票收益的关系甚至可能发生反转，因此大多数因子并不稳定。 进一步地，我们筛选本样本中对未来股票收益有显著解释力的因子，采用多空策略投资组合的未来收益最为异常现象强度的衡量指标，通过回归预测模型分析异常现象强度与市场情绪间的关系。实证结果表明，多数异常现象强度与市场情绪显著负相关。市场情绪高涨时，套利机会将更快地被理性投资者利用，错误定价持续的时间较短，因此异常现象强度较弱；市场情绪低落时，套利机会被理性投资者利用的速度降低，错误定价持续时间较长，因此异常现象强度较强。 市场情绪对套利机会被利用的速度的影响有两种可能的渠道，一种是客观流动性的约束限制了投资者套利的速度，另一种是投资者主观的套利意愿。我们在回归中加入流动性衡量指标发现，流动性的系数不显著，市场情绪指数系数没有明显变化。因此我们得出结论，市场情绪对异常现象强度的影响是由投资者情绪，而非流动性主导的，即市场情绪高涨时，市场交易活跃，投资者普遍更激进地利用套利机会，异常现象强度较低；市场情绪低落时，投资者对风险的承受能力有所降低，偏好安全的资产，因此异常现象强度较高。

According to traditional asset pricing theory, only systematic risk should earn risk premium, while idiosyncratic risk of securities can be diversified and therefore should not be priced. However, empirical study shows that return of market portfolio cannot fully explain return of securities. After adjusted with market return, those variables that still have significant explanatory powers for security returns are called factors, and this phenomenon is called anomaly. This paper selects 41 factors published in top finance journals to test with uniform samples. We construct long-short portfolios to check that whether these strategies generate significant returns. The results show that part of the factors have significant explanatory power for future stock returns. It can be inferred that the presence of factors is highly dependent on samples. Due to the existence of overreaction, the relationship may even be reversed, so factors are not stable.Furthermore, significant factors are selected from the sample. We adopt the return of long-short strategies as a measure of intensity of anomalies, then we regress return of long-short strategies on market sentiment. It is showed that they have significantly negative correlation. When the market sentiment is high, the arbitrage opportunity would be taken advantage by rational investors more quickly, so the intensity of anomalies is weak; when the market is depressed, the duration of anomalies would be longer, so the intensity of anomalies is strong.There are two possible channels for the impact of market sentiment on the speed at which arbitrage opportunities are used, constraints of liquidity or investors’ willingness. We add liquidity measure to the regression to find that the coefficient of liquidity is not significant and that the coefficient of market sentiment does not change significantly. Therefore, we conclude that the impact of market sentiment on the intensity of anomalies is driven by investors’ willingness, rather than liquidity.