Rank: Advanced Member
Groups: Registered, Registered Users, Unverified Users
Joined: 10/16/2009(UTC)
Posts: 34
|
Hello everyone,
i'm studying a special exponential moving average,described by R.F. Meyer in his "An adaptive Method of Routine Short Term Forecasting".Renato Di Lorenzo,in his recent book,has published the ProRealTime formula,but i can't translate it in Metastock language.
Can anyone help me?
Here is the ProRealTime code:
//Indicatore FiltroDiMeyer
datatime=N
PeriodoEma = 17
SmoothingEma= 2 / (PeriodoEma + 1)
PeriodoStdDeviationEma = 9
PeriodoStdDeviationPrezzo = 9
//condizioni iniziali
IF BarIndex=0 THEN
Ema = close
ELSE
//crea EMA di Meyer
Ema = Ema[1] * (1 - SmoothingEma) + close * SmoothingEma
SigmaEma = STD[PeriodoStdDeviationEma](Ema)
SigmaPrezzo = STD[PeriodoStdDeviationPrezzo](close)
Alfa = SigmaEma / SigmaPrezzo
SmoothingEma=Alfa
ENDIF
RETURN Ema[datatime] COLOURED(0,0,255) AS "Filtro di Meyer "
// Nella finestra 2-Variables si deve aggiungere la variabile N (con valore di default 0).
|
|
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Subscribers
Joined: 5/13/2005(UTC)
Posts: 715
Location: Midwest, USA
Was thanked: 1 time(s) in 1 post(s)
|
Well, this is what I came up with: Code:
{_FilterDiMeyer}
PeriodEma:=17;
SmoothingEma:=2/(PeriodEma+1);
PeriodStdDev:=9;
PeriodStdDevPrice:=9;
{ Initial Condition }
EMA:=If(Cum(1)=1,CLOSE,
{ DiMeyer EMA }
PREV*(1-SmoothingEma)+CLOSE*SmoothingEma);
{ Calc'd, but not plotted }
SigmaEma:=Stdev(EMA,PeriodStdDev);
SigmaPrice:=Stdev(CLOSE,PeriodStdDevPrice);
Alfa:=SigmaEma/SigmaPrice;
SmoothingEma:=Alfa;
{ Plot }
Ema;
But, it seems to plot IDENTICALLY to a 17 period Exponetial MA:
Are you sure you copied the code correctly? What software was it in, originally?
|
|
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Unverified Users
Joined: 10/16/2009(UTC)
Posts: 34
|
Thanks for your reply,jjstein.The code is correct and it works well on prorealtime.The problem seems to be all that you have calculated but not plotted.The classic ema formula is correctly :
EMA:=If(Cum(1)=1,CLOSE,
PREV*(1-SmoothingEma)+CLOSE*SmoothingEma);
but,in order to plot the desired ema,we would substitute SmoothingEma with Alfa.In this way we would obtain an adaptive ema,with adaptive factor equal to the ratio between the signal(Devstand of a classic ema in a certain period) end the noise(Devstand of the closes in the same period).
Forgive my poor english,i hope i was clear.
Regards
|
|
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Subscribers
Joined: 5/13/2005(UTC)
Posts: 715
Location: Midwest, USA
Was thanked: 1 time(s) in 1 post(s)
|
noumann wrote:Thanks for your reply,jjstein.The code is correct
and it works well on prorealtime.The problem seems to be all that you
have calculated but not plotted.The classic ema formula is correctly :
EMA:=If(Cum(1)=1,CLOSE,
PREV*(1-SmoothingEma)+CLOSE*SmoothingEma);
but,in order to plot the desired ema,we would substitute SmoothingEma
with Alfa.In this way we would obtain an adaptive ema,with adaptive
factor equal to the ratio between the signal(Devstand of a classic ema
in a certain period) end the noise(Devstand of the closes in the same
period).
Forgive my poor english,i hope i was clear.
Regards
If I understand you, then "Alfa" should be used to recalculate. the code should look like this: Code:
PeriodEma:=17;
SmoothingEma:=2/(PeriodEma+1);
PeriodStdDev:=9;
PeriodStdDevPrice:=9;
{ Initial Condition }
EMA:=If(Cum(1)=1,CLOSE,
{ DiMeyer EMA }
PREV*(1-SmoothingEma)+CLOSE*SmoothingEma);
SigmaEma:=Stdev(EMA,PeriodStdDev);
SigmaPrice:=Stdev(CLOSE,PeriodStdDevPrice);
Alfa:=SigmaEma/SigmaPrice;
SmoothingEma:=Alfa;
If(Cum(1)=1,CLOSE,
{ DiMeyer EMA }
PREV*(1-SmoothingEma)+CLOSE*SmoothingEma);
It might help if you could capture a screen from ProRealTime and put it here. Use PtrSc with Paint, save as PNG, upload to IMGUR and paste here. By any chance, can you post a link to the original article?
|
|
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Unverified Users
Joined: 10/16/2009(UTC)
Posts: 34
|
Here it is:
http://i.imgur.com/6YXJP.png?1
I also have a worksheet made by tha author to explain how to calculate the indicator.
Let me know how i can send you,
Thanks
|
|
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Unverified Users
Joined: 10/16/2009(UTC)
Posts: 34
|
|
|
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Subscribers
Joined: 5/13/2005(UTC)
Posts: 715
Location: Midwest, USA
Was thanked: 1 time(s) in 1 post(s)
|
Well, the plot from IMGUR is much smoother than mine, here:  But I don't know why. Maybe someone else can chime in.
|
|
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Subscribers, Unverified Users
Joined: 10/28/2004(UTC)
Posts: 3,111
Location: Perth, Western Australia
Was thanked: 16 time(s) in 16 post(s)
|
Code:
maPrd:=17;
sdPrd:=9;
data:=close;
ma:=extfml("forum.mov",data,maPrd,e);
StDevOfMA:=extfml("forum.stdev",ma,sdPrd,1);
StDevOfData:=extfml("forum.stdev",data,sdPrd,1);
ratio:=stdevOfMA/max(stdevOfData,0.00001);
meyer:=if(cum(isdefined(data+ratio))=1,data,prev*(1-ratio)+(data*ratio));
{plot}
meyer;
Requires the Forum.dll I'd have to dig into the source code of the forum.dll to see if it allowed fractional inputs to the ema in order to possible remove the PREV in the last line?/? wabbit [:D] [edit] There are going to be minor differences in the values for the initial 9 bars or so, as there are different seeds to the ema algos. After then though, the differences will be insignificant.
|
|
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Subscribers
Joined: 5/13/2005(UTC)
Posts: 715
Location: Midwest, USA
Was thanked: 1 time(s) in 1 post(s)
|
Wabbit: That gives identical results on 27 Apr 12 on the $COMP: 3030.56 -- Mine 3030.56 -- Wabbit 3029.4 -- Original at IMGUR.The "original" also has a much smoother curve.
|
|
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Subscribers, Unverified Users
Joined: 10/28/2004(UTC)
Posts: 3,111
Location: Perth, Western Australia
Was thanked: 16 time(s) in 16 post(s)
|
After a very quick play, I came up with another solution (and tested this time! [:)] ) which produces the same results as the spreadsheet... Code:
maPrd:=17;
sdPrd:=9;
data:=CLOSE;
ma:=ExtFml("forum.mov",data,maPrd,e);
StDevOfMA:=Stdev(ma,sdPrd);
StDevOfData:=Stdev(data,sdPrd);
ratio:=stdevOfMA/Max(stdevOfData,0.00001);
i:=Ref(Cum(IsDefined(data+ratio))=1,+1);
ratio:=ExtFml("forum.sum",ratio,1);
i:=ExtFml("forum.sum",i,1);
limit:=BarsSince(i=1)>=0;
meyer:=If(i,data,PREV*(1-ratio)+(data*ratio));
{plot}
limit*meyer
It uses a few tricks here and there to make things fit in the timescale, and although there is a forward Ref() it is not crystal balling prices. wabbit [:D]
|
|
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Subscribers
Joined: 7/25/2005(UTC)
Posts: 1,042
Was thanked: 57 time(s) in 54 post(s)
|
Hi Wabbit My version just focuses on removing both PREVs from Johnathan's version by using the MSTT.EMA DLL function. I doubt that the "code" boundaries will work (not if I believe the Preview) but I'll post it anyway and see what happens.. Code:
PeriodEma:=17;
SmoothingEma:=2/(PeriodEma+1);
PeriodStdDev:=9;
PeriodStdDevPrice:=9;
{ Initial Condition }
{ DiMeyer EMA }
EMA:=ExtFml("MSTT.EMA",C,PeriodEma);
SigmaEma:=Stdev(EMA,PeriodStdDev);
SigmaPrice:=Stdev(CLOSE,PeriodStdDevPrice);
Alfa:=SigmaEma/SigmaPrice;
SmoothingEma:=2/Alfa-1;
{ DiMeyer EMA }
ExtFml("MSTT.EMA",C,SmoothingEma);
Roy
|
|
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Subscribers, Unverified Users
Joined: 10/28/2004(UTC)
Posts: 3,111
Location: Perth, Western Australia
Was thanked: 16 time(s) in 16 post(s)
|
Hi Roy,
You and I know the MSTT.dll will do the trick (we wrote it) but I'm unsure of the distribution of the library.
Unfortunately, the Forum.dll statically casts the periods to their integer values, so it cannot be used to remove the PREV in this instance.
wabbit [:D]
|
|
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Subscribers
Joined: 5/13/2005(UTC)
Posts: 715
Location: Midwest, USA
Was thanked: 1 time(s) in 1 post(s)
|
The figure for 27 Apr 12 on the $COMP is still different: 3029.4 -- Original 3030.56 -- Mine 3030.56 -- Wabbit 3030.56 -- Roy The "original" has a smoother curve, as well.
ORIGINAL ------------  CURRENT -- Johnathan, Wabbit#2 and Roy ---------------------------------------------- 
|
|
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Unverified Users
Joined: 10/16/2009(UTC)
Posts: 34
|
thank you all...but i can't understand..may i use the wabbit's code?..and if yes,where can i find the dll?
|
|
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Subscribers
Joined: 5/13/2005(UTC)
Posts: 715
Location: Midwest, USA
Was thanked: 1 time(s) in 1 post(s)
|
noumann wrote:thank you all...but i can't understand..may i use the wabbit's code?..and if yes,where can i find the dll? The forum.dll is here. However, the three versions of the code produce identical results (slight difference in Feb 12 for Roy's version), and none match the original; something is still off...
|
|
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Subscribers, Unverified Users
Joined: 10/28/2004(UTC)
Posts: 3,111
Location: Perth, Western Australia
Was thanked: 16 time(s) in 16 post(s)
|
jjstein wrote:the three versions of the code produce identical results (slight difference in Feb 12 for Roy's version), and none match the original; something is still off..
Code:// FiltroDiMeyer
datatime=N
PeriodoEma = 17
SmoothingEma= 2 / (PeriodoEma + 1)
PeriodoStdDeviationEma = 9
PeriodoStdDeviationPrezzo = 9
//condizioni iniziali
IF BarIndex=0 THEN
Ema = close
ELSE
//crea EMA di Meyer
Ema = Ema[1] * (1 - SmoothingEma) + close * SmoothingEma
SigmaEma = STD[PeriodoStdDeviationEma](Ema)
SigmaPrezzo = STD[PeriodoStdDeviationPrezzo](close)
Alfa = SigmaEma / SigmaPrezzo
SmoothingEma=Alfa
ENDIF
RETURN Ema[datatime] COLOURED(0,0,255) AS "Filtro di Meyer "
// Nella finestra 2-Variables si deve aggiungere la variabile N (con valore di default 0).
In the OP code, there is the "datetime" variable which is assigned a value "N" and this never changes. There is no mention of an Ema[] function in the online help manual for the software, so I'm going to guess any differences lie in that Ema[] function; we might even discover that if what I suspect is happening, when we call Ema[0] that a default 14 or 21 period ema of close prices is returned? Who has the source code for that function? For those with access to the software, try changing the last line to:
Code:RETURN Ema[Alfa] COLOURED(0,0,255) AS "Filtro di Meyer "
and see what difference that makes?
wabbit [:D]
P.S. I will be home at the end of today and if not too jet lagged, may have another play, but I reckon we have the solution nailed, proving its better to have a description of how functions are supposed to work instead of other peoples interpreted code.
|
|
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Subscribers, Unverified Users
Joined: 10/28/2004(UTC)
Posts: 3,111
Location: Perth, Western Australia
Was thanked: 16 time(s) in 16 post(s)
|
Well...
the original code doesn't even compile in the application, so what chance have we of making something from it?
Thanks to the OP for wasting our time. [:(]
wabbit [:D]
|
|
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Subscribers
Joined: 5/13/2005(UTC)
Posts: 715
Location: Midwest, USA
Was thanked: 1 time(s) in 1 post(s)
|
wabbit wrote:the original code doesn't even compile in the application, so what chance have we of making something from it? Noumann -- Would you please check the original code, and/or post a link to the original? Make sure to use the CODE tags (see below), so nothing gets left out. Code:
//Indicatore FiltroDiMeyer
datatime=N
PeriodoEma = 17
SmoothingEma= 2 / (PeriodoEma + 1)
PeriodoStdDeviationEma = 9
PeriodoStdDeviationPrezzo = 9
//condizioni iniziali
IF BarIndex=0 THEN
Ema = close
ELSE
//crea EMA di Meyer
Ema = Ema[1] * (1 - SmoothingEma) + close * SmoothingEma
SigmaEma = STD[PeriodoStdDeviationEma](Ema)
SigmaPrezzo = STD[PeriodoStdDeviationPrezzo](close)
Alfa = SigmaEma / SigmaPrezzo
SmoothingEma=Alfa
ENDIF
RETURN Ema[datatime] COLOURED(0,0,255) AS "Filtro di Meyer "
// Nella finestra 2-Variables si deve aggiungere la variabile N (con valore di default 0).
|
|
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Unverified Users
Joined: 10/16/2009(UTC)
Posts: 34
|
|
|
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Subscribers
Joined: 5/13/2005(UTC)
Posts: 715
Location: Midwest, USA
Was thanked: 1 time(s) in 1 post(s)
|
Well, I tried plugging that in, but it doesn't work, either. Where did you get the original chart, posted to IMGUR? I wonder if that last line means to return an EMA-smoothed version...
|
|
|
|
|
|
| Users browsing this topic |
|
Guest (Hidden)
|
Forum Jump
You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.
