OpenEuphoria

Re: More help: Exponential Moving Averages

• Posted by Pete Lomax <petelomax at blueyonder.co.uk> Jan 19, 2003
• 413 views

On Sat, 18 Jan 2003 21:37:52 +0930, noranross at optusnet.com.au wrote:

>Thanks for Query The period is 5. The actual EMA is as follows. My
>calculations do not get these figures or anything close.
>Date    EMA    Close
>2/1/02   3419     3440
>3/1/02   3418     3414
>4/1/02   3425.5  3443
>7/1/02   3429    3463
>8/1/02   3436    3430
>9/1/02   3431   3420
>
>Ross Elliott Darwin

The following code does not always get the right results either,
actually, it only gets one answer right....  Maybe Yahoo Finance (or
where ever you got those figures from) does not use a textbook
formula, at the very least it looks to be rounding to nearest 0.5.  I
tried three variations, see what you think.

constant ema=3D{
{"2/1/02",	 3419,	  3440},
{"3/1/02",	 3418,	  3414},
{"4/1/02",	 3425.5,  3443},
{"7/1/02",	 3429,	 3463},
{"8/1/02",	 3436,	 3430},
{"9/1/02",	 3431,	3420}}

atom pma,csp
for i =3D 2 to 6 do
	pma=3Dema[i-1][2]
	csp=3Dema[i][3]
	-- print ema from above table and three attempts at calculating it
	printf(1,"%f:%f:%f:%f\n",{ema[i][2],
							(1/5)*csp+(1-1/5)*pma,
							(2/5)*(csp-pma)+pma,
							(2/6)*csp+(1-2/6)*pma})
-- a failed attempt at reverse engineering the  weighting:
--	printf(1,"                 %f/%f=3D%f\n",{(ema[i][2]-csp),(pma-csp),
--							     (ema[i][2]-csp)/(pma-csp)})
=09
end for
if getc(0) then end if

Pete