Research On Profit Forecast Of Steel Listed Companies

According to the data of "my steel" website,China’s steel output accounts for more than half of the global steel output in 2019,and the output value of the steel industry accounts for a high proportion,which plays an important role in the development of the national economy.Among them,36 steel and iron companies are currently listed in China’s A-share market,with a market value of more than 750 billion,which is also an important part of the Chinese market.Therefore,in this context,the development of the iron and steel company is closely concerned by investors and managers.At the same time,the profit forecast is a prerequisite for enterprise value evaluation.Therefore,through the profit forecast of the iron and steel company,the value of the company can be evaluated;However,due to the cyclical characteristics of the iron and steel company,the profit fluctuation is large,so the applicability of the conventional method to predict the profit is low.At present,there are mainly linear regression model and grey prediction model for profit prediction of iron and steel companies,but the prediction accuracy and applicability of these two methods can not meet the requirements of profit prediction: regression analysis method is limited by the number of samples and data,if more influencing factors are included,it will lead to large errors in model prediction;The grey prediction model has low timeliness and is not suitable for the long-term profit prediction of iron and steel companies.Therefore,this paper proposes to establish a BP neural network model based on total factor productivity,which can comprehensively consider the factors affecting profits.Total factor productivity is selected because of its characteristics.A single index can be used to synthesize many influencing factors and systematically reflect the micro characteristics of steel listed companies.After calculating the total factor productivity of iron and steel company,taking it as the independent variable of the model can significantly reduce the model error,make the prediction result close to the target value,which is conducive to the construction of the model.BP neural network model is selected because it is a non-linear model,and has the ability of self-learning.It can quickly calculate the results without assumptions.Compared with other models,the process of establishing BP neural network model is simpler,and the relationship between each independent variable can be considered.Therefore,this paper uses BP neural network method based on total factor productivity to forecast the profit of iron and steel companies,which can not only meet the market demand,but also make up for the lack of applicability of conventional methods.Through the analysis of BP neural network and the calculation of total factor productivity,with the help of anaconda,Matlab and Stata software,the data of steel listed companies in CSMAR,"my steel" network and National Bureau of statistics are extracted.After the calculation and modeling process,the BP neural network model based on total factor productivity is established.This paper establishes the model based on the data of 2019.In the aspect of performance test,the model has high accuracy and universality.It can predict the profit of iron and steel companies.The error of the profit forecast result is small,and the winning rate is equal to the analyst’s forecast result.It is helpful for external investors to judge the company’s value estimation.