Taking some form of moving averages yields a smoothing of time series which is delayed. However, taking moving averages in the reverse time direction gives a smoothing which is in advance. The two resulting smoothed time series are pointwise averaged getting as result a smoothed version with "no delay".
