The  classical  wavelet  methods  suffering  from  boundary  problems caused  by  the  application  of  the  wavelet  transformations  to  a  finite signal,  to  treatment  boundar

y  problems  with  wavelet  regression,  we propose  a  simple  method  that  decreasing  bias  at  the  boundaries,  it  is based on a combination of wavelet functions and local linear quantile

regression  (WR LLQ).  We  use  the  proposed  technique  to  forecast

stock index t ime series. Detailed experiments are implemented for the proposed method, in which WR LLQ, WR, and WRLP methods are compared.  The  proposed  WRLLQ   model   is  determined  to  besuperior  to  the  WR  and  WRLP  methods  in  predicting  the  stock