摘要
针对国家对证券市场的监管和投资人对证券市场的决策而言,一个精确度更高的时间序列预测模型是很重要的。本文在吸取前人利用神经网络模型进行证券价格趋势预测中的经验,基于长短期记忆人工神经网络搭建时间序列,提出利用指数移动平均值均线对数据进行采样,以提高时间序列对证券价格趋势预测的准确性。本文采用的时间序列数据为上证指数日线收盘价,实验结果表明预测精确度提高70%,采用指数移动平均值均线对数据进行采样输入时间序列分析模型的预测效果优于传统方法进行数据采样。本文的实验结果为国家对市场的监督和投资者的决策提供了一定参考。
In light of the government’s supervision of bond markets and investor’s decision of bond markets,a more precise time series model is very important.This paper has learned lessons of time series price trend prediction using LSTM neural networks of previous researchers and purposed a sampling method using exponential moving average for improving precision of time series model’s prediction of security price trend based on the construction of time series by using long-and short-term memory artificial neural network.This paper uses daily close price of Shanghai Securities Composite Index as data of time series.The result shows the precision of prediction increases by 70%and sampling based on exponential moving average is better than sampling based on traditional way.The result of this paper provides guidance for the government’s supervision and investor’s decision of bond markets.
作者
陈腾劲
CHEN Tengjin(Nanfang College Guangzhou)
出处
《中国商论》
2021年第20期92-94,共3页
China Journal of Commerce
