This page last changed on Sep 06, 2008 by

The wavelet filter smooths a great deal of the noise in the tick data but it leaves some bumps that make life difficult for a model. This page shows the result for the wavelet filter combined with an exponential moving average filter (EMA). As the graphs below show, the EMA produces a lot of the smoothing that the wavelet produces. The wavelet plus the EMA is better than the EMA alone, but the difference is not dramatic. When the wavelet is combined with the EMA, the EMA windows is 20 points wide.

Linear wavelet over 64 points

red market tick data
blue wavelet and EMA
magenta EMA over 24 points

Linear wavelet + EMA

Linear wavelet + EMA

wavelet_ema.jpg (image/jpeg)
wavelet_ema2.jpg (image/jpeg)