Seasonality Analysis Indicator small bug fix

This topic has 13 replies, 4 voices, and was last updated 6 years ago by Vonasi.

Viewing 14 posts - 1 through 14 (of 14 total)

01/08/2018 at 3:38 PM #57755

Vonasi

There is a very minor graphical error in the coding of my Seasonality Analysis indicator in the library. Please replace lines 256 to 302 with the following or alternatively download the corrected ITF file attached to this post.

IF StartMonth = 1 THEN
DrawText("            From Jan #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)
ENDIF

IF StartMonth = 2 THEN
DrawText("            From Feb #StartYear#" ,barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)
ENDIF

IF StartMonth = 3 THEN
DrawText("            From Mar #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)
ENDIF

IF StartMonth = 4 THEN
DrawText("            From Apr #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)
ENDIF

IF StartMonth = 5 THEN
DrawText("            From May #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)
ENDIF

IF StartMonth = 6 THEN
DrawText("            From Jun #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)
ENDIF

IF StartMonth = 7 THEN
DrawText("            From Jul #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)
ENDIF

IF StartMonth = 8 THEN
DrawText("           From Aug #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)
ENDIF

IF StartMonth = 9 THEN
DrawText("            From Sep #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)
ENDIF

IF StartMonth = 10 THEN
DrawText("            From Oct #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)
ENDIF

IF StartMonth = 11 THEN
DrawText("            From Nov #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)
ENDIF

IF StartMonth = 12 THEN
DrawText("            From Dec #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)
ENDIF

IF StartMonth = 1 THEN

DrawText(" From Jan #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)

ENDIF

IF StartMonth = 2 THEN

DrawText(" From Feb #StartYear#" ,barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)

ENDIF

IF StartMonth = 3 THEN

DrawText(" From Mar #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)

ENDIF

IF StartMonth = 4 THEN

DrawText(" From Apr #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)

ENDIF

IF StartMonth = 5 THEN

DrawText(" From May #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)

ENDIF

IF StartMonth = 6 THEN

DrawText(" From Jun #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)

ENDIF

IF StartMonth = 7 THEN

DrawText(" From Jul #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)

ENDIF

IF StartMonth = 8 THEN

DrawText(" From Aug #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)

ENDIF

IF StartMonth = 9 THEN

DrawText(" From Sep #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)

ENDIF

IF StartMonth = 10 THEN

DrawText(" From Oct #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)

ENDIF

IF StartMonth = 11 THEN

DrawText(" From Nov #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)

ENDIF

IF StartMonth = 12 THEN

DrawText(" From Dec #StartYear#",barindex,15.5,SansSerif,Standard,10)coloured(0,0,0)

ENDIF

3 users thanked author for this post.

Despair,

Nicolas,

GraHal

Vonasi

Moderator

Topics: 192

Replies: 5088

Been thanked: 1811 times

01/08/2018 at 5:06 PM #57781

Nicolas

Thank you, your indicator has been updated in the library. Great post! 🙂

Seasonality Analysis

Nicolas

Keymaster

Topics: 81

Replies: 20863

Been thanked: 4773 times

01/08/2018 at 5:23 PM #57784

Vonasi

Thanks Nicolas.

I have just converted the indicator into two new versions. One that looks back at a fixed number of years from today so that the latest results are always from the latest sample period. A sort of always up to date walk forward.

In the second version you can set an in sample and an out of sample period so as to be able to check reliability of results more accurately.

Once I’ve double checked that they do what they say on the tin do you think it is best that I post them to the library or maybe just put them in a related forum post. I’m a bit conscious of filling up the library with similar indicators as this may not be what you want for the library.

Vonasi

Moderator

Topics: 192

Replies: 5088

Been thanked: 1811 times

01/08/2018 at 5:31 PM #57786

Nicolas

Vonasi wrote:

I’m a bit conscious of filling up the library with similar indicators as this may not be what you want for the library.

Don’t worry about this, post them. Since indicators are linked with their tags, if someone want something related to “seasonality”, he/she will find a lot of things and that’s a good point.

Vonasi wrote:

In the second version you can set an in sample and an out of sample period so as to be able to check reliability of results more accurately.

OMG, sounds enormous! 🙂

Nicolas

Keymaster

Topics: 81

Replies: 20863

Been thanked: 4773 times

01/08/2018 at 6:15 PM #57796

Vonasi

Nicolas wrote:

OMG, sounds enormous!

Only 772 very repetitive lines of code. Thank goodness for cut and paste and the find and replace all facility!

1 user thanked author for this post.

Nobody

Vonasi

Moderator

Topics: 192

Replies: 5088

Been thanked: 1811 times

01/08/2018 at 7:31 PM #57802

Nicolas

You could have used external indicators as functions, instead of copy paste them X times in the same code.

Nicolas

Keymaster

Topics: 81

Replies: 20863

Been thanked: 4773 times

01/08/2018 at 9:24 PM #57810

Vonasi

I did originally set it up as two separate codes. One for in sample and the other for out but it felt a bit clumsy so I combined the two. I hate calling on external indicators. I much prefer stand alone code. Sometimes I just like to make my life difficult. It makes the beer taste better.

The code is now up to 939 lines as I’ve added a rating system for the combined in sample and out of sample returns for each month!

I may have to stop now or I will go to sleep dreaming of Jan, Feb, Mar…….

Vonasi

Moderator

Topics: 192

Replies: 5088

Been thanked: 1811 times

01/08/2018 at 9:43 PM #57813

Nicolas

😆 good job! Passion feeds passion. Have a good night.

Nicolas

Keymaster

Topics: 81

Replies: 20863

Been thanked: 4773 times

01/08/2018 at 9:58 PM #57814

Vonasi

Minor bug fix no.2!

I just noticed that the description when you hover over the speech bubble in the corner for the Seasonality Analysis indicator is incorrect as it was for an older version.

I’ve attached a new itf with the correct text.

Hopefully this is the last bug I spot.

1 user thanked author for this post.

Yannick

Vonasi

Moderator

Topics: 192

Replies: 5088

Been thanked: 1811 times

01/08/2018 at 11:40 PM #57822

Alai-n

(Post off topic folder)

Good evening Vonasi, you seem to me to have a good command of your subject. That’s why I would like to ask you if you would feel able to translate the attached code (easylanguage) into probuilder language?

Waves of Wolfe

//Wolfe Wave Pattern --------------------------- 
Pa:=Trough(4,LOW,5); Pb:=Peak(3,HIGH,5); P1:=Trough(3,LOW,5); P2:=Peak(2,HIGH,5); P3:=Trough(2,LOW,5); P4:=Peak(1,HIGH,5); P5:=Trough(1,LOW,5); tb:=TroughBars(1,LOW,5); eki1:=If(tb<=4 AND p1 > pa AND pb < p2 AND p1 < p4 AND p1 > p3 AND p3 > p5 AND p4 < p2,1,0); p1:=Trough(3,LOW,5); 
p3:=Trough(2,LOW,5); p5:=Trough(1,LOW,5); o1:=p1-p3; o2:=p1-p5; a1:=TroughBars(3,LOW,5)-TroughBars(2,LOW,5); a2:=TroughBars(3,LOW,5)-TroughBars(1,LOW,5); tan1:=o1/a1; 
tan2:=o2/a2; eki2:=Abs(tan1-tan2); If(eki1 AND eki2 <= .2 ,2,0 

//Codice Easy language ==============HEADER============== 
//Study based on Connors & Raschke's "Wolfe Waves" 
//====== DECLARATIONS ==== 
INPUTS: TICKCHG(10), {Change (number of ticks) required to set up a new swing high/low.} PCTCHG(.2), {Percent change in price to set up a new swing high/low} OCCUR(1), PLOTBARS(40), ETALINE(TRUE), ZIGZAG(TRUE); 
ARRAY: PBAR[50](0), P[50](0); 
VARS: X(0), PP(1), JA_SLOPE1(0), JLA_CTR(0), JLA_LINE(0), JAIRHBAR(0), { LAST SWING HIGH BAR} JAIRLBAR(0), {LAST SWING LOW BAR} LOWSEEK(FALSE), {LOOKING FOR A LOW OR A HIGH?} W(0), {COUNTER} 
JLA_IRH(0), {LAST SWING HIGH VALUE} JLA_IRL(99999), {LAST SWING LOW VALUE} JA_SLOPE2(0), JA_SLOPE3(0), 
JLA_PT1(0), WOLFE(0); 
{===========MAIN PROGRAM============} 

IF CURRENTBAR = 1 THEN P[50] = C; 
IF LOWSEEK = FALSE AND P[50] <= H THEN BEGIN P[50] = H; PBAR[50] = 0; 
END; 
IF LOWSEEK = TRUE AND P[50] >= L THEN BEGIN P[50] = L; PBAR[50] = 0; 
END; 
IF (LOWSEEK = FALSE AND PBAR[50] <> 0) THEN BEGIN IF (TICKCHG = 0 AND L < P[50] * ( 1 - PCTCHG / 100) ) 
OR (TICKCHG <> 0 AND L < ( P[50] - tickchg * minmove points)) THEN BEGIN IF ZIGZAG = TRUE THEN PLOT4[PBAR[50]](P[50],"SWINGS"); 
LOWSEEK = TRUE; 
FOR W = 1 TO 49 BEGIN PBAR[W] = PBAR[W+1]; P[W] = P[W+1]; 
END; 
P[50] = L; PBAR[50] = 0; END; END; 
IF (LOWSEEK = TRUE AND PBAR[50] <> 0) THEN BEGIN IF (TICKCHG = 0 

AND H> P[50] * ( 1 + PCTCHG / 100)) OR (TICKCHG <> 0 AND H > (P[50] + tickchg * minmove points )) THEN BEGIN IF ZIGZAG = TRUE THEN PLOT4[PBAR[50]](P[50],"SWINGS"); LOWSEEK = FALSE; 
FOR W = 1 TO 49 BEGIN PBAR[W] = PBAR[W+1]; P[W] = P[W+1]; 
END; 
P[50] = H; PBAR[50] = 0; END; END; 
IF TIME = LASTCALCTIME AND DATE = LASTCALCDATE AND P[48 - PP] <> 0 THEN BEGIN PP = -1; WOLFE = 0; WHILE WOLFE < OCCUR AND PP < 46 BEGIN PP = PP + 1; VALUE1 = P[47-PP]; VALUE2 = P[48-PP]; VALUE3 = P[49-PP]; VALUE4 = P[50-PP]; 
CONDITION1 = VALUE2 > VALUE1 AND VALUE4 > VALUE3 AND VALUE4 < VALUE2 AND VALUE3 < VALUE1 AND VALUE4 > VALUE1; 

CONDITION2 = VALUE2 < VALUE1 AND VALUE4 < VALUE3 AND VALUE4 > VALUE2 AND VALUE3 > VALUE1 AND VALUE4 < VALUE1; 
IF CONDITION1 OR CONDITION2 THEN WOLFE = WOLFE + 1; END; 
JA_SLOPE1 = (P[49-PP] - P[47-PP]) / (PBAR[47-PP] - PBAR[49-PP]); JA_SLOPE2 = (P[50-PP] - P[47-PP]) / (PBAR[47-PP] - PBAR[50-PP]); 
{LINE 1-3} VALUE90 = TL_New(DATE[PBAR[47-PP]],TIME[PBAR[47-PP]],P[47-PP], DATE[PBAR[49-PP]], TIME[PBAR[49-PP]],P[49-PP]); Value14=TL_SetColor(VALUE90, 5); {VALUE93 = TL_SetExtRight(VALUE90,TRUE);} VALUE94 = TL_SETEND(VALUE90,DATE[PBAR[49-PP]-PLOTBARS],TIME[PBAR[49-PP]- PLOTBARS],TL_GETVALUE(VALUE90,DATE[PBAR[49-PP]-PLOTBARS],TIME[PBAR[49-PP]-PLOTBARS])); 
{LINE 1-4} VALUE91 = TL_NEW(DATE[PBAR[47-PP]],TIME[PBAR[47-PP]],P[47-PP], DATE[PBAR[50-PP]], TIME[PBAR[50-PP]],P[50-PP]); Value14=TL_SetColor(VALUE91, 16); TL_SETEND(VALUE91,DATE[PBAR[49-PP]-PLOTBARS],TIME[PBAR[49-PP]- PLOTBARS],TL_GETVALUE(VALUE91,DATE[PBAR[49-PP]-PLOTBARS],TIME[PBAR[49-PP]-PLOTBARS])); 
{ETA LINE} 
IF ETALINE THEN BEGIN VALUE92 = TL_NEW(DATE[PBAR[48-PP]],TIME[PBAR[48-PP]],P[48-PP], DATE[PBAR[50-PP]], TIME[PBAR[50-PP]],P[50-PP]); Value14=TL_SetColor(VALUE92, 4); TL_SETEND(VALUE92,DATE[PBAR[50-PP]-PLOTBARS],TIME[PBAR[50-PP]- PLOTBARS],TL_GETVALUE(VALUE92,DATE[PBAR[50-PP]-PLOTBARS],TIME[PBAR[50-PP]-PLOTBARS])); 

END; END; 
IF DATE = LASTCALCDATE AND TIME = LASTCALCTIME AND ZIGZAG = TRUE THEN BEGIN 
JA_SLOPE3 = (P[50] - P[49]) / (PBAR[49] - PBAR[50]); FOR JLA_CTR = PBAR[49] DOWNTO PBAR[50] BEGIN PLOT4[JLA_CTR](P[49] + (PBAR[49] - JLA_CTR) * JA_SLOPE3,"Swings"); END; 
END; 
FOR W = 1 TO 50 BEGIN PBAR[W] = PBAR[W]+1; END;

//Wolfe Wave Pattern ---------------------------

Pa:=Trough(4,LOW,5); Pb:=Peak(3,HIGH,5); P1:=Trough(3,LOW,5); P2:=Peak(2,HIGH,5); P3:=Trough(2,LOW,5); P4:=Peak(1,HIGH,5); P5:=Trough(1,LOW,5); tb:=TroughBars(1,LOW,5); eki1:=If(tb<=4 AND p1 > pa AND pb < p2 AND p1 < p4 AND p1 > p3 AND p3 > p5 AND p4 < p2,1,0); p1:=Trough(3,LOW,5);

p3:=Trough(2,LOW,5); p5:=Trough(1,LOW,5); o1:=p1-p3; o2:=p1-p5; a1:=TroughBars(3,LOW,5)-TroughBars(2,LOW,5); a2:=TroughBars(3,LOW,5)-TroughBars(1,LOW,5); tan1:=o1/a1;

tan2:=o2/a2; eki2:=Abs(tan1-tan2); If(eki1 AND eki2 <= .2 ,2,0

//Codice Easy language ==============HEADER==============

//Study based on Connors & Raschke's "Wolfe Waves"

//====== DECLARATIONS ====

INPUTS: TICKCHG(10), {Change (number of ticks) required to set up a new swing high/low.} PCTCHG(.2), {Percent change in price to set up a new swing high/low} OCCUR(1), PLOTBARS(40), ETALINE(TRUE), ZIGZAG(TRUE);

ARRAY: PBAR[50](0), P[50](0);

VARS: X(0), PP(1), JA_SLOPE1(0), JLA_CTR(0), JLA_LINE(0), JAIRHBAR(0), { LAST SWING HIGH BAR} JAIRLBAR(0), {LAST SWING LOW BAR} LOWSEEK(FALSE), {LOOKING FOR A LOW OR A HIGH?} W(0), {COUNTER}

JLA_IRH(0), {LAST SWING HIGH VALUE} JLA_IRL(99999), {LAST SWING LOW VALUE} JA_SLOPE2(0), JA_SLOPE3(0),

JLA_PT1(0), WOLFE(0);

{===========MAIN PROGRAM============}

IF CURRENTBAR = 1 THEN P[50] = C;

IF LOWSEEK = FALSE AND P[50] <= H THEN BEGIN P[50] = H; PBAR[50] = 0;

END;

IF LOWSEEK = TRUE AND P[50] >= L THEN BEGIN P[50] = L; PBAR[50] = 0;

END;

IF (LOWSEEK = FALSE AND PBAR[50] <> 0) THEN BEGIN IF (TICKCHG = 0 AND L < P[50] * ( 1 - PCTCHG / 100) )

OR (TICKCHG <> 0 AND L < ( P[50] - tickchg * minmove points)) THEN BEGIN IF ZIGZAG = TRUE THEN PLOT4[PBAR[50]](P[50],"SWINGS");

LOWSEEK = TRUE;

FOR W = 1 TO 49 BEGIN PBAR[W] = PBAR[W+1]; P[W] = P[W+1];

END;

P[50] = L; PBAR[50] = 0; END; END;

IF (LOWSEEK = TRUE AND PBAR[50] <> 0) THEN BEGIN IF (TICKCHG = 0

AND H> P[50] * ( 1 + PCTCHG / 100)) OR (TICKCHG <> 0 AND H > (P[50] + tickchg * minmove points )) THEN BEGIN IF ZIGZAG = TRUE THEN PLOT4[PBAR[50]](P[50],"SWINGS"); LOWSEEK = FALSE;

FOR W = 1 TO 49 BEGIN PBAR[W] = PBAR[W+1]; P[W] = P[W+1];

END;

P[50] = H; PBAR[50] = 0; END; END;

IF TIME = LASTCALCTIME AND DATE = LASTCALCDATE AND P[48 - PP] <> 0 THEN BEGIN PP = -1; WOLFE = 0; WHILE WOLFE < OCCUR AND PP < 46 BEGIN PP = PP + 1; VALUE1 = P[47-PP]; VALUE2 = P[48-PP]; VALUE3 = P[49-PP]; VALUE4 = P[50-PP];

CONDITION1 = VALUE2 > VALUE1 AND VALUE4 > VALUE3 AND VALUE4 < VALUE2 AND VALUE3 < VALUE1 AND VALUE4 > VALUE1;

CONDITION2 = VALUE2 < VALUE1 AND VALUE4 < VALUE3 AND VALUE4 > VALUE2 AND VALUE3 > VALUE1 AND VALUE4 < VALUE1;

IF CONDITION1 OR CONDITION2 THEN WOLFE = WOLFE + 1; END;

JA_SLOPE1 = (P[49-PP] - P[47-PP]) / (PBAR[47-PP] - PBAR[49-PP]); JA_SLOPE2 = (P[50-PP] - P[47-PP]) / (PBAR[47-PP] - PBAR[50-PP]);

{LINE 1-3} VALUE90 = TL_New(DATE[PBAR[47-PP]],TIME[PBAR[47-PP]],P[47-PP], DATE[PBAR[49-PP]], TIME[PBAR[49-PP]],P[49-PP]); Value14=TL_SetColor(VALUE90, 5); {VALUE93 = TL_SetExtRight(VALUE90,TRUE);} VALUE94 = TL_SETEND(VALUE90,DATE[PBAR[49-PP]-PLOTBARS],TIME[PBAR[49-PP]- PLOTBARS],TL_GETVALUE(VALUE90,DATE[PBAR[49-PP]-PLOTBARS],TIME[PBAR[49-PP]-PLOTBARS]));

{LINE 1-4} VALUE91 = TL_NEW(DATE[PBAR[47-PP]],TIME[PBAR[47-PP]],P[47-PP], DATE[PBAR[50-PP]], TIME[PBAR[50-PP]],P[50-PP]); Value14=TL_SetColor(VALUE91, 16); TL_SETEND(VALUE91,DATE[PBAR[49-PP]-PLOTBARS],TIME[PBAR[49-PP]- PLOTBARS],TL_GETVALUE(VALUE91,DATE[PBAR[49-PP]-PLOTBARS],TIME[PBAR[49-PP]-PLOTBARS]));

{ETA LINE}

IF ETALINE THEN BEGIN VALUE92 = TL_NEW(DATE[PBAR[48-PP]],TIME[PBAR[48-PP]],P[48-PP], DATE[PBAR[50-PP]], TIME[PBAR[50-PP]],P[50-PP]); Value14=TL_SetColor(VALUE92, 4); TL_SETEND(VALUE92,DATE[PBAR[50-PP]-PLOTBARS],TIME[PBAR[50-PP]- PLOTBARS],TL_GETVALUE(VALUE92,DATE[PBAR[50-PP]-PLOTBARS],TIME[PBAR[50-PP]-PLOTBARS]));

END; END;

IF DATE = LASTCALCDATE AND TIME = LASTCALCTIME AND ZIGZAG = TRUE THEN BEGIN

JA_SLOPE3 = (P[50] - P[49]) / (PBAR[49] - PBAR[50]); FOR JLA_CTR = PBAR[49] DOWNTO PBAR[50] BEGIN PLOT4[JLA_CTR](P[49] + (PBAR[49] - JLA_CTR) * JA_SLOPE3,"Swings"); END;

END;

FOR W = 1 TO 50 BEGIN PBAR[W] = PBAR[W]+1; END;

To tell the truth I do not know what this code is worth! But currently, I think it is the only available free access.
I’ll be curious to see it work, but unfortunately I do not have the skills to translate it under Probuilder.
The basic idea is to see if its operation can be closer to that on the attached video …!

https://www.youtube.com/watch?v=nCCeLo6UsFk

Alai-n

Participant

Topics: 44

Replies: 223

Been thanked: 29 times

01/08/2018 at 11:54 PM #57826

Vonasi

Sorry Alai-n EasyLanguage is not my thing.

Thank you for thinking that I might be able to help but to be totally honest I dabbled with programming when I was a kid (in the days of ZX Spectrums and Commodore 64’s!) and I went to IT college after school while working for a company that wrote programs for city traders – but this was back in the late 80’s when just getting a line on a graph took a month. I gave it up quickly when I realised that I hated sitting behind a desk and went off and became an engineer instead. I have always loved finding solutions to problems which I guess is how I now end up coding PRT in my free time (which I have a lot of). I like creating indicators that can be adapted into strategies rather than just using off the shelf indicators. I only really started on PRT coding eight months ago so I am sure there are others on here who can be of more use. Sorry to disappoint.

On a separate note I spent some time studying Waves of Wolfe and to be honest with you found no merit in it. If you spend long enough staring at enough lines on enough graphs you can see what ever you want to see!

Vonasi

Moderator

Topics: 192

Replies: 5088

Been thanked: 1811 times

01/09/2018 at 2:24 PM #58614

Alai-n

Thank you for your reply Vonasi and too bad for my code Waves of Wolfe, which leaves me still very curious;)

In any case, thank you very much for your contribution to the effort of this site.

Alai-n

Participant

Topics: 44

Replies: 223

Been thanked: 29 times

01/10/2018 at 9:50 PM #58924

Yannick

Thanks for sharing your seasonality analysis.

Would you be able to mix your seasonality indicator and your trend checker to built seasonal trading algorithm?

Something inspired by http://systemtradersuccess.com/seasonal-strategies-built-right-way/ even there is no code published

Regards

Yannick

Participant

Topics: 13

Replies: 106

Been thanked: 28 times

01/10/2018 at 11:06 PM #58936

Vonasi

I’m not sure what the exact thing is that you are asking for?

I have read that article before you posted the link and I was inspired by the suggestion that walk forward analysis was better than starting with a fixed date and always using that fixed date each month. Better to move the start point forward one month every month. I converted my indicator into this a couple of days ago and then had a better idea. Why not have two periods that walk forward – an in sample and out of sample for easy comparison of reliability? So I converted it to do this. I then added a rating system for each month using the results from each sample weighted more on the more recent out of sample results. I have posted this to the library and it is pending review before being listed.

I like the idea in the article of looking at the historical results for the up and coming days and would like to write this as an indicator.

As for converting it all to a trading strategy that would take quite a bit of thought and I’m not sure that just trading now because this day last year for the last few years was profitable is a good basis for a strategy. Best just to use the analysis and maybe the rating score to adjust trade direction and position sizing month by month to give you an edge in other strategies.

I have had a lot of time recently for indicator writing but this will shortly change due to other commitments and I doubt I will have much time to code. I have a long list of ‘would like to do’s’ already and not enough hours in the day!

Vonasi

Moderator

Topics: 192

Replies: 5088

Been thanked: 1811 times

Author

Posts

Viewing 14 posts - 1 through 14 (of 14 total)

Find exclusive trading pro-tools on