基於打分法之多因子選股策略建構：中國 A股市場實證分析

Abstract

本研究以2006年至2020年的中國A股市場爲對象構建打分法下的多因子選股策略，首先以中國A股市場股灾發生前的數據爲樣本，在規模類、估值類、質量類、成長類等常見的10大類因子中選擇了18個因子對其有效性進行識別，經過橫斷面t檢定、IC-IR檢定法、相關性檢定、因子分層回測一系列步驟對各類因子進行有效性、穩健型、獨立性以及一致性進行分析，最終篩選出6個風格因子，分別爲總市值對數、賬面市值比、每股收益成長率、非流動性指標、過去一個月股價報酬、過去一個月BETA值。\n識別出有效因子後，再基於打分法對六因子進行隨機組合，發現在單因子的基礎上不斷加入其餘因子可以在一定程度上提升選股效果，樣本內績效最佳的投資策略爲由六因子共同排序打分所構建的策略，投組的夏普比率與信息比率分別達到了1.1767以及1.3061。爲檢驗六因子打分法的穩定性，將該策略應用於股灾後的數據進行回測檢驗，雖然結果顯示投組在整體上的表現仍優於基準指數滬深300，但在2017-2018年間的超額收益有所下降，最後透過投組歸因分析發現由於因子風格轉變以及行業輪動等原因造成策略發生階段性的失效。

This paper takes China`s A-share market from 2006 to 2020 as the object to construct a multi-factor investing strategy under the scoring method. First, taking the data before the 2015 China stock market crash as the sample, selected 18 factors from 10 common categories, such as size, valuation, quality, and growth to identify their effectiveness. Through a series of steps of cross-sectional T-test, IC-IR test, correlation test, and factor stratification test, the effectiveness, robustness, independence, and consistency of various factors were analyzed, and finally, 6 style factors were screened out: logarithm of total market capitalization, book-to-market ratio, EPS growth rate, Amihud ILLIQ, a stock price return of the past month, BETA of the past month.\nAfter the effective factors are identified, these six factors are randomly combined based on the scoring method. It is found that adding other factors based on a single factor can improve the effect of stock selection to a certain extent. The investment strategy with the best performance in the sample is the six-factor model, the Sharpe ratio and Information ratio of the portfolio reached 1.1767 and 1.3061. This strategy was applied to the market after the 2015 crash to test the stability of the six-factor scoring method. Although the results showed that the portfolio`s overall performance was still better than the benchmark index CSI 300, the excess return declined from 2017 to 2018. Finally, through the attribution analysis, we can find that the strategy fails periodically due to factor style and industry rotation change.