RE: MACD derivative

["Samerski, Michael (Australia)" <msamerski@xxxxxxxxxxxxxxxxxx>]

You want the period to be small so that it is sensitive the vol. changes BUT
you know that SD needs at least 30 samples (see your statistics text) to be
meaningful so I use 30.

BTW - I gave up on MS6.0 and didn't upgrade to 6.5. I now write my own
system testing software.

-----Original Message-----
From: Lionel and Gail Issen [mailto:lissen@xxxxxxxxxxxxxxxx]
Sent: 24 March 1999 10:25
To: metastock@xxxxxxxxxxxxx
Subject: Re: MACD derivative


Michael:

Thanks for this idea.

when I tried to put this into an indicator, the formula checker kept asking
for a bracket after the function name, like stdev(function, periods).This
of course resulted in an odd number of left and right brackets.  I finally
settled for the term "indicator".   Now my formula reads function( )/
stdev(indicator,periods).   I'll have to check and see if it is giving
valid answers.

Do you have any suggestions on the number of periods to use?

Lionel
-----Original Message-----
From: Samerski, Michael (Australia) <msamerski@xxxxxxxxxxxxxxxxxx>
To: 'metastock@xxxxxxxxxxxxx' <metastock@xxxxxxxxxxxxx>
Date: Tuesday, March 23, 1999 4:16 PM
Subject: RE: MACD derivative


>Here's a trick that I like.
>
>I find that the optimum trigger levels for most indicators are not really
>constant over time so I "normalise" the indicator before determining the
>trigger levels.
>I simply divide by the standard deviation.
>
>i.e. to normalise MACD
>
>NormMACD = MACD / SD(MACD)
>
>Try this on your favourite indicator e.g.
>
>NormRSI = RSI / SD(RSI)
>
>Michael Samerski.