|
I take Preston saying "...and would love to see a shorter
/ easier version of it" to be an invitation to take the discussion
further.
Actually wabbit himself in his post
has nicely dissected the recursive averaging to its well approximated simpler
version as below;
"...it might be interesting to note that the
AverageOfMovingAverages (the mathematical average of the ten 2 bar SMAs) is
ALMOST the same as a much more simple _expression_, Mov(C,6,S). If you
compare the PRECISE VALUES of the AverageOfMovingAverages and the Mov(C,6,S)
there is always a small difference, but, if you compare the instances when the
CLOSE crosses the AverageOfMovingAverages and the instances when the CLOSE
crosses the Mov(C,6,S) they are the same, with about 3-4% error. If you
apply one bar latitude in either direction, the two expressions are the same
within 1%. Thefore, for testing when the CLOSE crosses the
AverageOfMovingAverages the trader could easily substitute Mov(C,6,S) for the
more complicated _expression_."
But what I was more interested in
RMO was not the formula in itself which when the indicator itself is available
has no more additional use but how it, so well, tackled the 'gaps' or the
wildness of a couple of ticks in the direction opposite to the trend / position.
Most usual MACO system would have created a lag and if a signal had been
generated in that skew it would have carried on for quite a while but was not so
in RMO. When the whole Rainbow Indicator formula itself is taken for studying,
the process does not become obvious but when the simplified version of wabbit is
considered it makes eminent sense.
What better way than to average the
skewedness of a couple of ticks with more saner ones prior or past to them to
reduce the impact of this skew. Simple averaging of essentially a short period
makes sure equal weightage is given to the saner ones regardles of their
positioning - whether before or after the 'gaps' / the sudden spurts
thereby reducing the impact of this few stray behaviour of the market while
still in a larger trend. Then the resultant output can always be used
for long period averaging to make sure one sits through the trend
inspite of these few stray & adverse ticks. In hindsight, it all looks so
very simple & logical. I seriously wonder whether the the designer of RMO
himself realised it, for if he had, he could very well have gone for the long
period exponential averaging of the simple moving averaging like Mov( Mov(C,6,S)
, 81 , E ) instead of choosing to average the Rainbow Indicator thereby
losing some amount of original thinking.
While wild moves of very
short term in nature is ignored, the adverse effect of this would be a much more
severe lag because of the initial simple averaging. In other words, this sytem
while avoiding smaller and sharper strayness would either get into the trend
later but by which time the probability of trend having set in would have become
high. By same logic, it would also get out of the trend later. Or
take bigger loses / bigger whipsaws when prices trade in larger ranges
due to its lack of sensitivity. That is, while avoiding smaller whipsaws it will
take larger ones (though they may be fewer) but also lose good amount of profits
at the time of exits even when in trend which explains the words of Big Papa
"..For all the testing of the RMO
I have done, it is good at getting in, but terrible at getting
out.."
The limitation of any Moving
Average System has probably been best described by Preston...
"If the lag is removed then there are more whipsaws. If the
whipsaws
are dampened, then the moving average is later to the party. There
is
only so much information that can be extracted from price and
volume
data no matter how many ways it is tortured, twisted and
manipulated."
Must thank everybody who contributed for a good learning period for
me.
----- Original Message -----
Sent: Friday, February 20, 2009 4:01 AM
Subject: [EquisMetaStock Group] Re: adjusted moving averages
&zerolagoscillators
> GV,
>
> Today you have learned the formula
to the RMO/Rainbow and that
> programmers never lay all there cards on
the table.
>
> I'd say you've learned quite a bit.
>
>
I actually like the recursive moving average and would love to see a
>
shorter / easier version of it.
>
>
Preston
__._,_.___
![]()
__,_._,___
|