Denoising of financial time series using wavelet analysis

Subject Areas : Financial engineering

Hojjatollah Sadeghi ¹ (Assistant Professor University of Yazd, Department of Accounting and Finance)
zahra dehghani firoozabadi ² (Master of Financial Engineering and Risk Financ, Yazd University of ‎Science and Arts)

Received: 2017-03-14 Accepted : 2018-01-13 Published : 2017-09-23

Keywords: Time series, Wavelet Transform, noise, Wavelet analysis, denoising,

Abstract :

every series of Wavelet coefficients includes part of time series in the scale of different time series. Implementation of the wavelet transform, using   the best Wavelet at the right levels has significant impact on the results of the results of the financial analysis. the purpose of  this study is to explanation of the importance of the concept of scale-time and the use of different  time intervals in  checking the behavior of the financial markets to be determined whether the removing noise from the  time series can  accurate the  decisions we have to make in the future or not? Therefore we analyzed 16 selected index of the Tehran Stock Exchange using software "R" and using Wavelet transformation up to five levels for 250 data then put them all under noise removing process. In the next step we used two methods for evaluation the noise removing process. one clustering all the selected index in the dendrogram method And the other one time series predictions of total index which includes 500 data and  the use of the data that has been noise removed  into two methods of Haar wavelet and Daubechies. The results of both method claim better performance using Wavelet removing noise using Daubechies wavelet in this series. our main goal is  using the wavelet analysis and noise removing from  time series and  using that  in financial topics.