Possible Trendline Bug

Questions about MultiCharts and user contributed studies.
User avatar
ABC
Posts: 518
Joined: 16 Dec 2006
Has thanked: 110 times
Been thanked: 293 times
Contact:

Possible Trendline Bug

Postby ABC » 24 Aug 2010

I stumbled across some weird Trendline behavior in Multicharts current release, where the changing of the Bar Spacing changes the direction of a Trendline. I could reproduce it with a simple demonstration code, which I'll attach to the post.
This could be a huge problem if you rely on Trendlines in your trading, so any comment is highly appreciated.

Picture 1 is showing the Trendline correctly drawn:
Image

Picture 2 is the same Trendline with the bar spacing decremented once more - what was up is down now:
Image

A down Trendline will show similar behaviour, just vice versa:
Image

What was down becomes up:
Image

Code used to demonstrate the issue:

Code: Select all

Inputs:
   BarNum      (1000);
   
Var:
   TLID      (-1);
   
if CurrentBar >= BarNum and TLID = -1 then begin
   if High[2] < High then begin
      TLID = TL_New_S(Date[2], Time_S[2], High[2], Date, Time_S, High);
      TL_SetColor(TLID, green);
      TL_SetExtRight(TLID, true);
   end
   else
   if Low[2] > Low then begin
      TLID = TL_New_S(Date[2], Time_S[2], Low[2], Date, Time_S, Low);
      TL_SetColor(TLID, red);
      TL_SetExtRight(TLID, true);      
   end;
end;


If you reduce the bar spacing further, the trendline will flip up and down.

Regards,
ABC

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 51 times

Re: Possible Trendline Bug

Postby Tresor » 24 Aug 2010

Confirmed on my end.

User avatar
ABC
Posts: 518
Joined: 16 Dec 2006
Has thanked: 110 times
Been thanked: 293 times
Contact:

Re: Possible Trendline Bug

Postby ABC » 24 Aug 2010

Thanks Tresor, I appreciate that.

User avatar
Anastassia
Posts: 179
Joined: 18 Jan 2010
Been thanked: 4 times

Re: Possible Trendline Bug

Postby Anastassia » 24 Aug 2010

Hi Guys,

Modify technique as follows please. Change coordinates around.

Code: Select all

Inputs:
   BarNum      (1000);
   
Var:
   TLID      (-1);
   
if CurrentBar >= BarNum and TLID = -1 then begin
   if High[2] < High then begin
      TLID = TL_New_S(Date, Time_S, High, Date[2], Time_S[2], High[2]);
      TL_SetColor(TLID, green);
      TL_SetExtRight(TLID, true);
   end
   else
   if Low[2] > Low then begin
      TLID = TL_New_S(Date, Time_S, Low, Date[2], Time_S[2], Low[2]);
      TL_SetColor(TLID, red);
      TL_SetExtRight(TLID, true);     
   end;
end;


Depending on the coordinates "extend to the right" could be both, move up and move down.

Thank you

User avatar
ABC
Posts: 518
Joined: 16 Dec 2006
Has thanked: 110 times
Been thanked: 293 times
Contact:

Re: Possible Trendline Bug

Postby ABC » 24 Aug 2010

Anastassia,

thank you, but this is just meant as a workaround until the issue is fixed, correct?

Besides that your suggestion creates another issue, instead of flipping the direction, now the study can change the Trendline angle when you decrease the bar spacing. Which is also crucial.

Thank you,
ABC

bowlesj3
Posts: 2021
Joined: 21 Jul 2007
Has thanked: 174 times
Been thanked: 402 times

Re: Possible Trendline Bug

Postby bowlesj3 » 24 Aug 2010

Just curious, What happens if you create the trend line on currentbar = 1 (extended already and the correct color already but hidden at value zero and extending flat to the right), then use the MoveTrendLine_s function in the thread below to move it into position when the criteria are met. Although I do not extend my trend lines right (I use them for other uses only) I have only seem this function move a trend line incorrectly 2 times since I wrote it (touch wood). It also traps any bad moves now. I change bar squeeze a lot every day. So I suspect it will have no impact on this function. I have one use where I put out 301 TrendLines at the start of the study and they get moved all day with no problem. As far as I can tell, so far the only two bad moves I have seen with this function are ones where the trend line was perfectly vertical.

viewtopic.php?f=5&t=7598

User avatar
ABC
Posts: 518
Joined: 16 Dec 2006
Has thanked: 110 times
Been thanked: 293 times
Contact:

Re: Possible Trendline Bug

Postby ABC » 25 Aug 2010

bowlesj3,
I noticed the problem with trendlines created on the first bar and moved into position later. So I came up with this simple example to demonstrate the problem. The positioning works fine and the trendlines are correct, unless you decrease the bar spacing to a certain point.
This is how it looks with Anastassia's code modification:

The first picture showing the trendline as it should look like and then I decreased the bar spacing and took a picture as soon as the trendline changed. The code didn't recalculate inbetween nor was anything else changed than the bar spacing:

Image
Image
Image
Image
Image

With the coding example in the first post the trendlines would flip up and down instead of changing the angle, like they do in this example.

Regards,

ABC

bowlesj3
Posts: 2021
Joined: 21 Jul 2007
Has thanked: 174 times
Been thanked: 402 times

Re: Possible Trendline Bug

Postby bowlesj3 » 25 Aug 2010

The main question now is "did you try the function MoveTrendLine_s?"
It has a reversing code built into it using an if test.
I am not sure if it does automatically what Anastassia suggested. Maybe!

Anyway, my understanding is, regardless of whether the trend line was
1/ created at the desired location or whether it was
2/ later moved to that desired location
the trend line initially is position correct.
However after either of these two positioning methods has worked correctly, simply decreasing the bar spacing causes the trend line to shift to a vertical position.

So I would assume that if you added a 3rd method of initially placing the trend line correctly in position (that being manually placing the trend line correctly) then later squeezing the bars would create the same "undesired vertical positioning result". Is this assumption correct? If not what if you manually position the trend line in reverse order and extend it in the opposite direction to it extends to the right. What happens then?

So it may be that (since the function I wrote has the reversing if test built in to the function) that this in fact may be a "4th way of placing the trend line initially into correct position".

So maybe that is why my function on very rare instances screws up (where as my old methods of moving trend lines screwed up about 20 to 30 times a day). Maybe it is the bar spacing and the function has it mostly correct but not 100% correct for certain bar spacings. Very Interesting. Maybe I will put the current bar spacing into the function's built in trap print statement.

User avatar
Anastassia
Posts: 179
Joined: 18 Jan 2010
Been thanked: 4 times

Re: Possible Trendline Bug

Postby Anastassia » 25 Aug 2010

Hi ABC,

Besides that your suggestion creates another issue, instead of flipping the direction, now the study can change the Trendline angle when you decrease the bar spacing. Which is also crucial.

Angle changed in both cases. When decreasing the bar spacing trendline angle changes only visually. You can check that the change of angle is only visual effect by comparing coordinates before and after bar spacing decrease. Coordinates will be the same.
See attached screenshot please

Thank you
Attachments
Trendline.zip
(6.74 KiB) Downloaded 57 times

User avatar
ABC
Posts: 518
Joined: 16 Dec 2006
Has thanked: 110 times
Been thanked: 293 times
Contact:

Re: Possible Trendline Bug

Postby ABC » 25 Aug 2010

Hi Anastassia,

so you are saying everything is working as it should?
Then please explain why does the angle change when I decrease the changing and change back to normal when I decrease the spacing any further?
With your explanation - which I agree to, this is the way it should be and it should work - this shouldn't be happening then, correct? You can see it in the pictures I uploaded to this thread today, how the angle flips back and forth and doesn't decrease slowly. This just doesn't seem like it is working correct to me.

Thank you,
ABC

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 51 times

Re: Possible Trendline Bug

Postby Tresor » 25 Aug 2010

It is clearly a bug.
Attachments
TL bug.png
TL bug.png (20.76 KiB) Viewed 597 times

bowlesj3
Posts: 2021
Joined: 21 Jul 2007
Has thanked: 174 times
Been thanked: 402 times

Re: Possible Trendline Bug

Postby bowlesj3 » 25 Aug 2010

I agree. It is clearly a bug. this just should not be so hard. My function may have solved a different problem but there never should have been a need for it in the first place. This problem is ever stranger. The bottom line is when you create a trend line or move it is simply should go there and stay there period. MC should figure out any complexities and the user should not need to worry about it at all (and if they have to worry about it the buck stops with MC). Not only that as far as I am concerned MC's habit of (after lastbaronchart is reached) deleting and redrawing trend lines, text and arrows until barstatus=2 is also a bug.

User avatar
Anastassia
Posts: 179
Joined: 18 Jan 2010
Been thanked: 4 times

Re: Possible Trendline Bug

Postby Anastassia » 25 Aug 2010

Hi Guys,

Then please explain why does the angle change when I decrease the changing and change back to normal when I decrease the spacing any further?

ABC, please come to our live chat to illustrate this for us.
Thank you

It is clearly a bug.

Tresor,
When you draw red line please go to Format Window and set the same coordinates for it as it is for the green line. Then decrease the bar spacing again and you will see that it will behave same way.
Thank you

User avatar
ABC
Posts: 518
Joined: 16 Dec 2006
Has thanked: 110 times
Been thanked: 293 times
Contact:

Re: Possible Trendline Bug

Postby ABC » 25 Aug 2010

Anastassia,
there is everything outlined in this thread so you can exactly reproduce the issue, you just have to follow the simple steps described. I don't know how I could show it any clearer, even in the live chat.

Thank you,
ABC

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 51 times

Re: Possible Trendline Bug

Postby Tresor » 25 Aug 2010

Anastassia wrote:Then decrease the bar spacing again and you will see that it will behave same way.
Thank you


Then please take this little quiz attached.
Attachments
TL extension quiz.png
TL extension quiz.png (13.42 KiB) Viewed 598 times

User avatar
Anastassia
Posts: 179
Joined: 18 Jan 2010
Been thanked: 4 times

Re: Possible Trendline Bug

Postby Anastassia » 25 Aug 2010

ABC,

We don't see any issue on our end and tried to explain in this thread that this is the way it works which makes sense.
I am not sure what you meant by the following,
Then please explain why does the angle change when I decrease the changing and change back to normal when I decrease the spacing any further?


Thank you

User avatar
ABC
Posts: 518
Joined: 16 Dec 2006
Has thanked: 110 times
Been thanked: 293 times
Contact:

Re: Possible Trendline Bug

Postby ABC » 25 Aug 2010

Anastassia wrote:ABC,
I am not sure what you meant by the following,
Then please explain why does the angle change when I decrease the changing and change back to normal when I decrease the spacing any further?


Thank you


Anastassia,

please take a look at the four pictures I posted in the seventh post of this topic. You see how the trendline angle flips back and fourth although the bar spacing decreases from picture to picture.
These are the four pictures. Notice how the trendline changes it's angle while the bar spacing decreases from picture to picture. The bar spacing is the only thing changed, the study is not recalculated or anything.

Image
Image
Image
Image
Image

I don't understand how you can't see that this is an issue.

Thank you,
ABC
Last edited by ABC on 25 Aug 2010, edited 1 time in total.

User avatar
Anastassia
Posts: 179
Joined: 18 Jan 2010
Been thanked: 4 times

Re: Possible Trendline Bug

Postby Anastassia » 25 Aug 2010

Tresor,

Are you saying that you decreased bar spacing and trendline from vertical changed to diagonal?

Thank you

User avatar
Anastassia
Posts: 179
Joined: 18 Jan 2010
Been thanked: 4 times

Re: Possible Trendline Bug

Postby Anastassia » 25 Aug 2010

ABC,
We do see from the pictures that trendline flips with the decrease of bar spacing. However, we can't reproduce it on our end.
So please visit our live chat to show us.

Thank you in advance

User avatar
TJ
Posts: 6553
Joined: 29 Aug 2006
Location: Global Citizen
Has thanked: 966 times
Been thanked: 1893 times

Re: Possible Trendline Bug

Postby TJ » 25 Aug 2010

I have tested the code (post#1),
I clicked on the [><] button 3~4 times and could not make the trendline flip.

However the trendline will flip at extreme compression.
I had to click the [><] button ~7 times to make it flip.

When I checked the trendline property, I found the values were not changed.




p.s. I don't think Anastassia's code is the proper way to align trendlines. The bottom/left points are the beginning coordinates .

User avatar
ABC
Posts: 518
Joined: 16 Dec 2006
Has thanked: 110 times
Been thanked: 293 times
Contact:

Re: Possible Trendline Bug

Postby ABC » 25 Aug 2010

Thanks TJ. I was just trying to bring this to TS Supports attention and it's seems to be just a visual problem. But for someone relying on Support and Resistance trendlines this might be a huge issue in visual trading.

Anastassia,
please follow the steps TJ describes with your code, this should reproduce it. At least this seems to do it for the other posters in this thread.

Thank you,
ABC

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 51 times

Re: Possible Trendline Bug

Postby Tresor » 25 Aug 2010

Anastassia wrote:Tresor,

Are you saying that you decreased bar spacing and trendline from vertical changed to diagonal?

Thank you


ABC did it - I only interpreted his picture. And his trendline changed first from diagonal to vertical to diagonal to vertical, etc.

As a result the trendline extension is pointing to different places.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 51 times

Re: Possible Trendline Bug

Postby Tresor » 25 Aug 2010

Anastassia,

This series of screenshots will help you over identifying the problem.

Picture 1.

BTW, I can't attach more than 1 picture in 1 post (something must have been changed in the forum).
Attachments
TL1.png
TL1.png (20.99 KiB) Viewed 589 times

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 51 times

Re: Possible Trendline Bug

Postby Tresor » 25 Aug 2010

Picture 2
Attachments
TL2.png
TL2.png (19.4 KiB) Viewed 585 times

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 51 times

Re: Possible Trendline Bug

Postby Tresor » 25 Aug 2010

The last one.

Picture 3
Attachments
TL3.png
TL3.png (20.6 KiB) Viewed 588 times

bowlesj3
Posts: 2021
Joined: 21 Jul 2007
Has thanked: 174 times
Been thanked: 402 times

Re: Possible Trendline Bug

Postby bowlesj3 » 25 Aug 2010

I tried squeezing bars (so much so I can't see the trend lines). I can't get any of my trend lines to flip.
1/ I have tons of them (most are on 1 minute bars).
2/ I move most of them. (created on currentbar = 1). Many using MoveTrendLine functions.
3/ Bar sizes are 10second, 1min,5min,10min,15min,30min,60min.
4/ Using MC 6.0 beta2.
5/ Using IB TWS feed.
NOTE: I do not have 2 minute bars.

Yesterday I had 1 minute bars mismatch the 5 minute bars which is not suppose to happen. A reload fixed it. Might be worth a try.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 51 times

Re: Possible Trendline Bug

Postby Tresor » 25 Aug 2010

bowlesj3 wrote:I tried squeezing bars (so much so I can't see the trend lines). I can't get any of my trend lines to flip.


Lucky one :)

bowlesj3 wrote:NOTE: I do not have 2 minute bars.


The trendline flipping / inaccuracy phenomena is irrespective of resolution (I observe it on any timeframe 1 minute, 10 minutes, 1 hour, etc.).

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 51 times

Re: Possible Trendline Bug

Postby Tresor » 25 Aug 2010

ABC,

You might want to look into this thread in which I discussed possible bug in bar spacing / bar thickness: viewtopic.php?f=1&t=7117

User avatar
ABC
Posts: 518
Joined: 16 Dec 2006
Has thanked: 110 times
Been thanked: 293 times
Contact:

Re: Possible Trendline Bug

Postby ABC » 26 Aug 2010

Tresor,
your three pictures in this thread show exactly how serious the problem is.
When you decrease the bar spacing, of course the visual appearance of two trendlines would change, but the cross still needs to be found at the same date, time and price. This needs to be the case, as the slope of the trendlines can't change and neither do the start points - both can't be affected by changing the bar spacing.
Conclusion: There is something wrong here.

Thank you,
ABC

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 51 times

Re: Possible Trendline Bug

Postby Tresor » 26 Aug 2010

ABC wrote:When you decrease the bar spacing, of course the visual appearance of two trendlines would change, but the cross still needs to be found at the same date, time and price.


Exactly. The cross point should stay always in the same price and time.

ABC wrote:There is something wrong here.


1. This is a very serious bug, at least for me as I only trade trendlines.

2. There is something wrong with the forum. Why was this topic put in PL Programming and User contributed Studies, and not in Charts?- this looks like a charting problem.

bowlesj3
Posts: 2021
Joined: 21 Jul 2007
Has thanked: 174 times
Been thanked: 402 times

Re: Possible Trendline Bug

Postby bowlesj3 » 26 Aug 2010

Hi ABC,

For sure. TSS knows for sure. Someone needs to get online with them live as it is happening (as they asked for). That is the only way they can figure it out. That is why I created my post above because it is not happening to me and of course not happening to most users. If it was happening to most users they would be able to recreate the problem and fix it very very fast. When they can not recreate it they can not fix it. So the bottom line is if you want it fixed get on line with them. That way they can dig inside your system and get some debugging data to fix it.

John.

User avatar
ABC
Posts: 518
Joined: 16 Dec 2006
Has thanked: 110 times
Been thanked: 293 times
Contact:

Re: Possible Trendline Bug

Postby ABC » 26 Aug 2010

John,
out of curiosity, did you test it with the code posted in this thread? I stumbled upon this problem with a study that uses a multiple of trendlines and within this study maybe one or two of the trendlines were affected by this issue. With the demo code it happens all the time, so my guess is that the trenlines are more prone to show this strange behavior, if start and end point are close to each other.

I agree btw that this thread is located in the wrong forum, it's not an issue with power language in my opinion.

Regards,
ABC

bowlesj3
Posts: 2021
Joined: 21 Jul 2007
Has thanked: 174 times
Been thanked: 402 times

Re: Possible Trendline Bug

Postby bowlesj3 » 26 Aug 2010

Hi ABC,

Actually no I didn't (pressed for time these day). I looked over the code to see if I could figure out any problems with it (relative to that function I wrote) but I could not see any problems. The fact that the anchors for the trend line were close together is a good thing to point out to them. I just checked it again and your points are 2 bars apart (that is very close for a trend line) The fact that MC is messing up when the coordinates are very close is very interesting actually because it relates back to another problem I have noticed. I suspect that MC is reading the screen somehow to do some of its work and if the bars are too close it can't get it right. My problem was in my binary search if I remember. I think I submitted a valid time and it could not find the bar for it even though I could see it. Very rare. I will have to try and find that problem to get a refresher on the details.

John.
Last edited by bowlesj3 on 27 Aug 2010, edited 1 time in total.

User avatar
Anastassia
Posts: 179
Joined: 18 Jan 2010
Been thanked: 4 times

Re: Possible Trendline Bug

Postby Anastassia » 27 Aug 2010

Hi ABC,
Neither myself or developers are able to reproduce the issue on our end.
We would really appreciate if you could visit our live chat to show us the trenline flipping.
Thank you

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 51 times

Re: Possible Trendline Bug

Postby Tresor » 27 Aug 2010

I came to the live chat and Anastassia was able to reproduce the issue remotely on my PC (picture 1, picture 2 and picture 3 from one of my previous posts).

MC developers say it is not a bug... In my opinion it is a bug.

Anastassia will for sure present MC developers' arguments for not classing it as a bug.

ABC, what now?

User avatar
Anastassia
Posts: 179
Joined: 18 Jan 2010
Been thanked: 4 times

Re: Possible Trendline Bug

Postby Anastassia » 27 Aug 2010

Hi Guys,

Thank you Tresor for taking your time to show us the trendline question via remote connection. It is very much appreciated!

We have analyzed the question with senior developers.
They concluded that at such high compression rates, the trend lines will not function properly (i.e. retain the direction of the trend that can be seen at lower compression rates).
This happens for the following reason: the trend lines connect the center points of the two end points. The actual points are not infinitely small, they have a defined size that is small enough to be considered high-definition charting. However, eventually even the smallest points rendered by our graphics engine start overlaying, being calculated as on top of each other. The relationship between center points of the two end points changes, and so does the direction of the trend line.
The expectation to have the trend lines retain the original direction on any compression rate is unrealistic. We are not aware of any other charting program that will be able to accurately maintain the relationship between points at such high compression rates. The current performance is suitable for most users.
Solution - to draw reliable trendlines, use significant points in the data series (such as major peaks and troughs), and also do not use such high compression.
The conclusion - this is not a bug, but a natural limitation, which is suitable to our product. Everyone is entitled to their own opinion, but a bug is performance that does not correspond to manufacturer’s specifications. This behavior is expected - it is a realistic limitation, and should be avoided with our recommendations, which are provided above.

Thank you

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 51 times

Re: Possible Trendline Bug

Postby Tresor » 27 Aug 2010

Anastassia wrote:They concluded that at such high compression rates, the trend lines will not function properly.

Anastassia wrote:The conclusion - this is not a bug.


Anastassia, I am sorry, but this is a self-detonating statement. More on the concept of self-detonating statements here:

http://www.youtube.com/watch?v=VeKys6zog2M
http://www.youtube.com/watch?v=z1M2CnEg ... re=channel


BTW, I highly recommend this Canadian philosopher. He is probably the most bright contemporary philosopher in the world, and the funniest one :) http://freedomainradio.com/


Anastassia wrote:The current performance is suitable for most users.

I wonder, if their Stochastics or RSIs or other studies showed different values depending on the compressions, would the most users be happy too?


Anastassia wrote:Solution - to draw reliable trendlines, use significant points in the data series (such as major peaks and troughs), and also do not use such high compression.

My solution:

1. Fix it - most preferable

2. Disable high compressions in MC - less preferable

3. Remove trendlines from drawing tools - least preferable

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 51 times

Re: Possible Trendline Bug

Postby Tresor » 27 Aug 2010

Anastassia wrote:This happens for the following reason: the trend lines connect the center points of the two end points. The actual points are not infinitely small, they have a defined size that is small enough to be considered high-definition charting. However, eventually even the smallest points rendered by our graphics engine start overlaying, being calculated as on top of each other. The relationship between center points of the two end points changes, and so does the direction of the trend line.


If the above were to make any sense then one could expect that the more compression is made the more errenous the trendlines will be. From what I observe the trendlines on highly compressed charts can sometimes be more accurate than on medium compressed charts - that's why you have flipping up and down.


Anastassia,

Please tell the senior developers that most probably the part of the engine that is responsible for compressing charts has bug(s).

If you compress a chart by let's say 40% then both (i) the bar width should be reduced by 40% and (ii) the bar spacing should be equally reduced by 40%. Or yet some other better logic should be applied to find the optimal relation between bar width and bar spacing when compressing / decompressing charts.

Picture with explanation attached.

BTW, unless this is fixed not only trendlines should be considered untrustworthy, but also a bunch of other drawing tools like:
- arcs
- equidistant channels
- time based fibbonaccis
- gann tools
- regression channels
- andrew's pitchfork
- others
Attachments
Compression engine.png
Compression engine.png (13.2 KiB) Viewed 589 times

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 51 times

Re: Possible Trendline Bug

Postby Tresor » 27 Aug 2010

Unless there is something I miss, part of MC engine that is responsible for compressing / decompressing charts is seriously flawed: bar spacing changes but the width of the bar does not. In my opinion both should dynamically adjust - otherwise there is little chance the trendlines will be correct.

I attach a picture to supplement my previous post.
Attachments
compression (il)logic.png
compression (il)logic.png (29.66 KiB) Viewed 587 times

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 51 times

Re: Possible Trendline Bug

Postby Tresor » 27 Aug 2010

I made a little excercise. I now have no doubt that the relation between bar spacing and bar width is being incorrectly displayed in MC as one decompresses charts.

This excercise is impossible to repeat with compressed charts.

Picture + explanation attached.

I think this is a logical incosistiency in charting. Maybe its purpose is to make charts readable / cleaner?

Regards
Attachments
MC incorrect bar spacing 1.png
MC incorrect bar spacing 1.png (39.46 KiB) Viewed 592 times

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 51 times

Re: Possible Trendline Bug

Postby Tresor » 27 Aug 2010

Last picture.
Attachments
Summary trendline.png
Summary trendline.png (35.66 KiB) Viewed 588 times


Return to “MultiCharts”