基于BP神经网络方法的量化投资——以券商为例
Quantitative Investment Based on BP Neural Network Method—Taking Securities Companies as an Example

本文先是对BP神经网络以及量化投资的历史进了简的解，再是对BP神经网络以及量化投资的原理做了基本的说明，并且对我国证券行业从几个宏观的角度做了分析。最后以R语言中的neuralnet函数建立了我国证券行业的研究模型。基于BP神经网络预测的可靠性，以我国证券板块为研究对象，从我国整个证券行业中选取年度财务数据，以其中554组数据作为BP神经网络的训练集，再结合中信证券2005年至2020年的16组真实的数据，使经过训练的BP神经网络模型对其进行预测，并且通过预测的数据与实际的数据对比，最终得到了较为可靠的模型，并根据此模型成功选取了10支股票，并以此提出了一些投资建议。
Firstly, this paper briefly explains the history of BP neural network and quantitative investment, then makes a basic explanation of the principle of BP neural network and quantitative investment, and analyzes China’s securities industry from several macro perspectives. Finally, the research model of China’s securities industry is established by using the neuralnet function in R language. Based on the reliability of BP neural network prediction, taking China’s securities sector as the research object, this paper selects the annual financial data from the whole securities industry in China, takes 554 groups of data as the training set of BP neural network, combined with 16 groups of real data of CITIC Securities from 2005 to 2020, makes the trained BP neural network model predict it, and compares the predicted data with the actual data. Finally, a more reliable model is obtained, 10 stocks are successfully selected according to this model, and some investment suggestions are put forward.