• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>International Joint Conference on Neural Networks >Financial Time Series Forecasting Using Non-Linear Methods and Stacked Autoencoders
【24h】

Financial Time Series Forecasting Using Non-Linear Methods and Stacked Autoencoders

机译：使用非线性方法和堆叠式自动编码器的金融时间序列预测

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Financial time series forecasting is considered to be one of the most challenging areas in current time series analysis theory. This complexity is caused mainly by the multitude of factors that influence the series and the volatility of this factors, which means that predicting models must adapt to a wide range of circumstances. Yet, the possibility of great financial return makes it a field of great interest for many researchers. Recently, non-linear techniques have been widely used for the construction of forecasting models, due to its ability to access higher order statistics. In particular, neural networks have shown to greatly enhance classical predicting models, and over the past years Deep Learning techniques are on the rise since computational power and data availability are becoming less of a constraint. This paper aims to compare the use of deep learning methods over shallow neural networks when predicting next day closing prices of stocks traded in the Brazilian financial market. Moreover, simulations using recent historical data are performed so that the models' profitability can be assessed and compared with the well-known strategy of Buy and Hold.
机译：金融时间序列预测被认为是当前时间序列分析理论中最具挑战性的地区之一。这种复杂性主要由影响该系列的众多因素和这种因素的波动，这意味着预测模型必须适应各种各样的情况。然而，伟大的财政回报的可能性使其成为许多研究人员非常兴趣的领域。最近，由于其访问更高阶统计数据的能力，非线性技术已被广泛用于预测模型的构建。特别是，神经网络已经显示出大大增强了经典预测模型，并且在过去几年中，深度学习技术正在上升，因为计算能力和数据可用性变得越来越少。本文旨在在预测巴西金融市场上交易股票的第二天闭合价格时，比较深入学习方法在浅层神经网络中的使用。此外，使用最近的历史数据进行模拟，以便可以评估模型的盈利能力，并将其与众所周知的购买策略进行比较。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号