Empirical Test Of Multi-factor Quantitative Stock Selection And Investor Sentiment Timing Strategy

Quantitative investment has a history of more than 40 years in foreign countries,but in China,quantitative investment has only really begun in the early 2000 s.The success of foreign quantitative funds represented by the Medallion Fund has attracted many investors to participate in researching and developing strategies of quantitative investment.In the sharp fluctuations of China's stock market in 2015,quantitative investment funds has been widely recognized by investors because of its low risk,and then quantitative investment strategy are constantly being developed.However,compared with foreign countries,China's quantitative investment is still in its infancy.At present,the main users of quantitative investment strategy are quantitative funds,and the most of the quantitative funds are adopted by the traditional multi-factor stock selection model.However,there are still many problems such as keeping the timeliness of the effective factor and selecting suitable time to carry out the strategy,there are still a lot of room for improvement.On the one hand,factors for selecting stocks have a certain timeliness,and their effectiveness need to be further verified in the new market environment.On the other hand,the multi-factor stock selection model lacks consideration of broad market risks while emphasizing the improvement of factors.As a result,most models have strong capability of selecting stocks,while being weak in timing selection.Therefore,this paper optimizes the timing of traditional multi-factor stock selection model.This article first uses the stocks in CSI300 as the sample stock pool,selects the monthly data of sample stocks from January 1,2009 to December 31,2019,and uses a third-party quantitative trading platform to conduct single-factor detection based on Python research modules.Specifically includes: introducing 30 factors as candidate factors,pre-processing the sample data,that is,outlier,standardization,and neutralization of market value and industries;screening effective factors through IC method,monotonicity test and correlation test.In order to improve the timeliness of the factor,this article uses the factor value of the 12 months before to select effective factors at the time of adjusting holding stocks.The position of stocks will be adjustedby quarterly,in other words,the validity of the factor will be maintained until the next quarter.Secondly,the scoring method was used to build a traditional multi-factor stock election model by equal weight,the CSI 300 is used as the benchmark portfolio,and the model was backtested in the backtest module of the ricequant platform.The level of risk and return about the strategy were compared and analyzed with the CSI300,it can be seen that the multi-factor stock selection model in this paper can obtain higher excess returns than the benchmark portfolio,but from the perspective of the risk indicators,the performance of the multi-factor stock selection model is slightly inferior to the benchmark portfolio,and the model still exists much room for optimization.In order to further optimize the model's return and risk indicators,this article attempts to construct an investor sentiment index,use this index to predict the market risk,and introduce it as a timing strategy into a multi-factor stock selection model.The specific steps include: referring to existing literature,selecting sentiment proxy indicators such as consumer confidence index and closed-end fund discount rate;using principal component analysis to construct market sentiment indexes,and further analyzes the relationship between the index and the month returns of CSI300.The data shows that the relationship between the change of the sentiment index and the returns of CSI300 is more significant.Therefore,this article uses the change of the sentiment index as the basis for the timing strategy.Considering that investors are more sensitive to the market's downward trend,therefore,if the positive change of the sentiment index appears for three consecutive periods,it is considered that investor sentiment is too high,and it is likely to reverse in the next period.If the negative change of the sentiment index appears for two consecutive periods,the investor sentiment is considered to be pessimistic and the market is under upward pressure.Therefore,this article evades investments in the third period with continuous positive values and the second period with continuous negative values,and introduces this timing condition into the traditional multi-factor selection.The result shows that the strategy of introducing investor sentiment index as the timing condition can significantly optimize the returns and risk indicators of the multi-factor stock selection strategy,and the indicators of excess returns,sharp ratios,and maximum drawdowns have been significantly optimized.It can be seen that the rolling test of factors effectively improves the stock selection ability of the multi-factor model,and the introduction of investor sentiment index to predict market risk also significantly improves the model's timing ability.