Patrick,
Views of experienced users like Roy are always valuable. However, here are some of my observations about EMA, WSM and their relationship. It is right that N period WSM is similar to (N*2)-1 Period EMA. Hence obviously, it starts early on the charts and we get NA plots during half of the duration than corrosponding EMA plots. However, during the early period of the plot when we have both EMA as well as WSM, the values are not same - they only start converging slowing, and ultimately get to be same, once there is enough data on the chart.
For example, if we use 14 period WSM and 27 period EMA - on the right side of the chart the value are same (I loaded 213 Daily bars of the sample image security). However, during 27 th to 140 th bars the values are not exactly the same. If we use the Forum DLL EMA function then it is right that during the inital seeding period the values will be same as Simple Moving Average, because both EMA and WSM will use simple moving average during this period - however we will expect the WSM to use SMA only for 14 bars and thereafter it will be using its own formula, whereas EMA will continue to use SMA upto 27th bar. So, the values will be different for the 14th to 26th bar also.
I have attached a chart, which should help put forth what I wanted to express. The top plot has been plotted with formula:
pds:=If(Cum(1)<14,Cum(1),14);
PREV+(1/pds*(C-PREV));
I have found that its value is always same as the inbuilt WSM - except during the initail seeding bars (When inbuilt WSM is not available) and during those bars, its value is rightly same as SMA (As suggested by Roy). I hope, we get the WSM function to work like this formula without the use of slow Prev function.
Regards
SMG