Research On Gem Sealing Investment Based On XGBoost Algorithm

GEM is also known as the second stock trading market,providing financing channels and growth spaces for entrepreneurial enterprises that do not meet the listing requirements of the motherboard,which has attracted many small and medium-sized entrepreneurial enterprises with a lower entry.Since the official listing in 2009,the size of the GEM market has been continuously expanded.The heat of investors is constantly rising.However,most of the circulation stocks in the GEM have generally existed the shortcomings of the market value,and how to choose more.Excellent GEM’s share of stocks has become a big problem that plagues investors.Quantitative investment combines statistical and financial knowledge to build investment strategy models,using computer implementation of programming transactions,is a convenient and efficient investment method,but its investment strategy is often dependent on the effectiveness of the model selected by the model.Fama and French have proposed a famous three-purpose model,and a large amount of effective factors have been explored by the market,but the effectiveness of the characteristic factors can have a major change in market environment,how to efficiently and accurately select effective factors to become an academic concern.Hotspots.The machine learning is widely applied in the multi-factor stocking model,so this paper is based on the more advanced XGboost algorithm for the core algorithm.Verify that the algorithm has a certain actual reference value.First of all,this paper studies the GEM,through analyzes the characteristics of recent GEM market,found that the market is used to use quantitative investment in the market with certain feasibility and necessity;secondly summarize the characteristics of the start of the GEM and compare the motherboard,find the GEM The stock has the characteristics of high premium,high volatility,so in short,active quantified investment methods in the GEM market.Secondly,this paper uses the SVM algorithm,the GBDT algorithm,and the XGBoost algorithm to build the stock model.The sample data set is the relevant data of the start of the GBDT algorithm and the AUC value of the GBDT algorithm and the XGBoost algorithm are found higher than 0.7,verify the machine.The feasibility of learning algorithms in the GEM Choice;compares the advantages and disadvantages and prediction results of each algorithm,and found that XGBoost’s actual reference value is higher,and the algorithm has a certain advantage in the startup collection stock.Finally,this paper constructs investment strategies based on the XGBoost algorithm in the GEM market,and the test sample is GEM stocks from January1,2020,to January 1,2022,and the interpretation capacity based on the stocking model is 5 days.10 days of investment strategy,found that the XGBoost algorithm has exceeded 1.10-day holding strategy,Sharp ratio exceeds 1.The accumulated rates reached 142.49%,190.93%,and the excess yields reached 61.19%,109.63%,The constructed investment strategy has a certain actual reference value in the GEM market;at the same time,based on the SVM and GBDT algorithm,we can construct a 10-day stock period strategy.It is found that its evaluation index is weak in XGBoost,so the XGBoost algorithm builds investment strategies in the GEM market.More practical.Based on the above analysis,this paper concludes that,first,invest in the GEM market can take short-term,active quantization investment method;Second,the XGBoost algorithm has strong advantages in solving the issue of GEM,and finally based on the XGBoost algorithm The short-term investment strategy of constructed has a certain actual reference value for guiding the GEM market.