I have written a test rig to help validate indicators. It is not generic, unfortunately; the indicator to be tested is only tested on its first argument. I wrote this to verify the EMA in a more complex setting; this is how I discovered the problem I mentioned below. There are instructions in the pod. Th. > -----Original Message----- > From: Weigert, Thomas > Sent: Friday, February 29, 2008 10:01 AM > To: Weigert, Thomas; 'devel AT geniustrader.org' > Subject: RE: [GT] New EMA and backup SEMA > > This time attached.... > > > -----Original Message----- > > From: Weigert, Thomas > > Sent: Friday, February 29, 2008 10:00 AM > > To: 'devel AT geniustrader.org' > > Subject: RE: [GT] New EMA and backup SEMA > > > > I should give a word of caution regarding the reimplemented EMA. > > > > While I tested it and compared it with TA_lib regarding output, and > found > > it to work, both when used inside of indicators and when used from the > > command line, I found one situation where it does not work as it should: > > > > When I use it inside of the SMI I recently wrote (a rather complex use > of > > nested indicators) I find that for some reason I don't understand the > SMA > > that starts of the EMA is not recorded properly, and so the EMA fails. I > > cannot see what causes that but am exploring. (If I change the > computation > > of the starting point to just use the current value rather than the n- > day > > SMA, then there are no problems.) Note that while having many other > > problems, the original EMA did not have this difficulty. > > > > I imagine it has to do with the nesting of indicators and dependencies > and > > the I:Generic:ByName, that somewhere some update gets lost. I will > explore > > further and hopefully uncover another important insight re GT which I am > > obviously still missing.... > > > > I am attaching SEMA.pm, which is the EMA starting with the actual value > of > > the data series. > > > > Th. > > > > > -----Original Message----- > > > From: Robert A. Schmied [mailto:ras AT acm.org] > > > Sent: Friday, February 29, 2008 1:38 AM > > > To: devel AT geniustrader.org > > > Subject: Re: [GT] New EMA > > > > > > > > > But currently > > > > that is what is happening.... I could try to make the EMA work like > > SMA > > > > as one need not do the check on the interval (it just uses the > > > > dependencies to compute the SMA for the starting value. > > > >
Attachment:
Test.pm
Description: Test.pm