因子投资是最常用的量化投资方法之一，因子是被量化的金融异象，有效的因子能够持续在金融市场产生相对于市场的超额收益。随着因子投资被越来越多地使用和研究，因子的普遍性质也被更多地探讨，包括因子动量。因子动量是因子收益保持前期状态的一种趋势，即“过去表现好的因子未来表现好，过去表现差的因子未来表现差”，这类似于长久以来被讨论的股票动量效应，但其研究目标从股票收益转变为因子收益。因子动量概念最初在2018年提出，而在A股仍未有对此的系统研究，因此本文对A股的因子动量效应进行实证检验。本文选取了常用的A股定价因子，使用分层法构建组合计算其收益序列，并对收益序列进行处理，剔除因子本身长期表现带来的时序自相关性。对处理后的因子收益的多项检验表明，A股在时序上和截面上都普遍存在因子动量效应，而时序因子动量强于截面因子动量。并且构建因子动量时存在明显的最佳形成窗口和滞后窗口参数，即使用滞后3个月的过去4个月收益之和能够在未来1个月上获得最优的预测效果。为验证因子动量在实际投资中的利用价值，本文首先根据因子动量对各因子收益直接组合，获得了相对于等权基准更高的收益。除此之外，本文在中证800股票池上构建投资组合，并使用因子动量对股票权重进行调整，也获得了相比于基准的显著超额收益，说明因子动量能够有效提升投资组合表现。最后，本文探讨了不同形成窗口参数下因子动量和股票动量间的关系，二者具有相关性，但分别在特定参数下相对对方具有超额收益，即并不能互相解释。

Factor investing is one of the most commonly used quantitative investment approaches. Factors are quantified financial anomalies and effective factors can continue to generate excess returns relative to the market. As factor investing is increasingly adopted and studied, some general properties of factors are also being explored, including factor momentum. Factor momentum is a trend that factor returns maintain the earlier state, that is, "factors with good performance in the past will perform well in the future, and factors with poor performance in the past will perform poorly in the future", which is similar to the stock momentum effect discussed for a long time, with study object changing from stock returns to factor returns. The concept of factor momentum was first proposed in 2018, and there is still no systematic research about it in Chinese A-share market. Therefore, this paper empirically studies the effect of factor momentum in A-share. commonly used A-share pricing factors are selected and their returns are calculated using the hierarchical method. The autocorrelations brought by the long-term performance of the factors are eliminated beforehand. Multiple tests of the treated factor returns show that factor momentum effect exists commonly in A share market, both directionally and cross-sectionally, with directional factor momentum being stronger. In addition, we find that there is a prominently optimal formation window regarding to factor momentum, that is, using the sum of returns in the past 4 months with a lag of 3 months can obtain the optimal prediction performance in the next month. In order to verify the practical value of factor momentum in actual investment, we firstly combine the returns of each factor directly according to the factor momentum, and obtain a higher return than the equal-weighted benchmark. In addition, this paper constructs a portfolio on the CSI800 stock pool, and uses factor momentum to adjust the stock weights, which again obtains significant excess return compared to the benchmark, showing factor momentum can effectively improve the performance of the investment portfolio. Finally, a spanning test reveals the relationship between factor momentum and stock momentum. They are correlated, but they have excess returns relative to each other under specific formation windows, indicating the difference between factor momentum and stock momentum.