Multicharts 6.0 Beta 3 feedback

Questions about MultiCharts and user contributed studies.
brendanh
Posts: 158
Joined: 07 Apr 2007
Has thanked: 1 time

Postby brendanh » 18 May 2010

Re shortcut keys:
1. Enable\disable automation (preferably different shortcut keys for each).
2. Buttons in dialogues when automation is exited. Simply underline a letter in the button text to denote the shortcut key as per normal Windows standards.
Dear bowlesj3 and brendanh,

We will give users the possibility to customize all commands in the next version of MultiCharts.

Dave Masalov
Thanks Dave, another position of mine was wrongly closed when IB exited for nightly maintenence tonight. Greater control over these options will be greatly appreciated.

User avatar
Dave Masalov
Posts: 1712
Joined: 16 Apr 2010
Has thanked: 51 times
Been thanked: 489 times

Postby Dave Masalov » 19 May 2010

Beta3 ONLY_ For the second day in a row on 2 different computers, continuous contracts won't update (ES with IB), they stop overnight.
Dear sptrader,

Our engineers have confirmed that it is a bug which will be fixed in the next version of MultiCharts 6.0

User avatar
geizer
Posts: 375
Joined: 16 Jun 2008
Has thanked: 40 times
Been thanked: 38 times

"Linked to All" does not work in some circumstance

Postby geizer » 19 May 2010

Dear TS Support,

Frankly this request/bug report was long due, but I was reluctant to send it because way too many times the response I receive from TSS is like: "the behavior is normal". Please investigate the current algorithm behind the "Linked to All" feature. The details are provided below:

"Link to All" doesn't work in the following circumstances:

1. Suppose we have a workspace with 4 chart windows each displaying a different instrument. For example: AAPL, INTC, GOOG, and NVDA. Symbol linking is set to "Not linked" for all of the above charts.

2. Next we create the 5th window with symbol IBM and set symbol linking to "Link to All". The purpose of this 5th window is to display an intraday resolution of the symbol plotted with with in some other window in daily/weekly resolution.

Problem/question: How can I force the new chart window (with "linked to all" activated) to be actually linked to any other window?

Things I tried / Is it a reasonable solution ?
I tried the following:
Made the first (AAPL) window active by mouse click, then typed AAPL in the command line - nothing happens. If I type some other instrument, example: CSCO, then both the 1st and my new 5th window will update and plot CSCO chart. Only after that if I type in the command line AAPL again, the symbol in the 5th window will be updated to AAPL. This is inconvenient, and demand unnecessary wiggling with the mouse - way too much. Even assuming this is a bug, which need to be fixed the "Linked to All" can be greatly enhanced as described below:

Suggestions / Improvements:

It appears a common sense to have "Linked to All" window behave as following:
- Force linking of the 5th window if symbol typed in the command line (EVEN IF SYMBOL IS THE SAME)
- Very Important: Link to the Active Chart Window by default - ALL THE TIME. In other words just link to whatever window the mouse pointer is placed into (truck the active chart window). This will allow automatic chart linking when reviewing charts. For example in a workspace with many smaller size daily resolution charts, the chart with intraday resolution and will be linked to all. This will allow to take a quick look at intraday charts by selecting windows with daily resolutions. By implementing the above the use of "Linked to All" feature will become much more valuable, and will not be limited to a scanner.

--
Thank you for considering,
Pavel

User avatar
Henrik
Posts: 140
Joined: 13 Apr 2010
Has thanked: 25 times
Been thanked: 11 times

Postby Henrik » 21 May 2010

Is there a way to see the live-performance (trades) of a strategy? I can't find such thing...?
View - Performance Report I think is what you are looking for.

I was wondering why the live results do not match with the Performance Report...
It seems that the performance report is only a simple backtest in the time of enable-strategy.
Scalping in Backtests is impossible. I cant so analyse the trades if the performance report based ob backtest, not on real trades you see in the position tracker.

sptrader
Posts: 742
Joined: 09 Apr 2010
Location: Texas
Has thanked: 483 times
Been thanked: 274 times
Contact:

Postby sptrader » 21 May 2010

Beta3 ONLY_ For the second day in a row on 2 different computers, continuous contracts won't update (ES with IB), they stop overnight.
Dear sptrader,

Our engineers have confirmed that it is a bug which will be fixed in the next version of MultiCharts 6.0
** Thanks Dave !

User avatar
Dave Masalov
Posts: 1712
Joined: 16 Apr 2010
Has thanked: 51 times
Been thanked: 489 times

Postby Dave Masalov » 24 May 2010

Re: "Scaling Issue"

I am sorry, but the behavior is NOT normal. You missing the point why I'm turning off title bars - because current implementation causes many errors when clicking on relatively small size windows. When title bars turned off, a new problem arises which is inability to maximize the chart window. The suggestion to make windows larger and break them into different work spaces is not suitable for me. There is a reason why I need so many windows visible at a time. Your response tells me TS Support is stubborn to accept the fact that "Title Bar auto-hide" implementation is half-done. Sure, denying the problem is much easier than attempting to find a solution to a problem. I guess that TSS spent less time discussing the problem than I spent time truing to find more efficient ways to work around these problems and wasting time for extra clicks and repairing trend lines that were accidentally displaced as a result of what you call "normal" behavior. Then multiply the time wasted by single user by a number of users.

Basically the way I interpret your (TSS) response to the problem I reported is "shut up, no one else reported the problem. We are not going to question our opinion on what's normal. We have more important problems to work on".
Dear Pavel,

Please do not exagerate things. We understand your point of view. And we do recognise there is a problem here. However it is not a five minutes fix and changing this behavior requires some very complex calculations to be done. It is not possible for the moment.

"Link to All" doesn't work in the following circumstances:

1. Suppose we have a workspace with 4 chart windows each displaying a different instrument. For example: AAPL, INTC, GOOG, and NVDA. Symbol linking is set to "Not linked" for all of the above charts.

2. Next we create the 5th window with symbol IBM and set symbol linking to "Link to All". The purpose of this 5th window is to display an intraday resolution of the symbol plotted with with in some other window in daily/weekly resolution.

Problem/question: How can I force the new chart window (with "linked to all" activated) to be actually linked to any other window?
The "Link to All" feature depends on changes made to other charts. However, when you type the same symbol name in the command line, you do not actually change anything.

We understand that it may cause inconveniences and plan to change this behavior in the future. We will consider your suggestions. Thank you.

** Thanks Dave !
Dear sptrader,

You are welcome.

User avatar
geizer
Posts: 375
Joined: 16 Jun 2008
Has thanked: 40 times
Been thanked: 38 times

Postby geizer » 25 May 2010

Dear Pavel,

Please do not exagerate things. We understand your point of view. And we do recognise there is a problem here. However it is not a five minutes fix and changing this behavior requires some very complex calculations to be done. It is not possible for the moment.
...
The "Link to All" feature depends on changes made to other charts. However, when you type the same symbol name in the command line, you do not actually change anything.

We understand that it may cause inconveniences and plan to change this behavior in the future. We will consider your suggestions. Thank you.
Dave, Thank you for support.

I entirely understand the time constraints and rely on TS Support expertise to consider implementing suggested improvements when time and opportunity allows.

User avatar
Dave Masalov
Posts: 1712
Joined: 16 Apr 2010
Has thanked: 51 times
Been thanked: 489 times

Postby Dave Masalov » 28 May 2010

Holidays option help to filter incorrect data, if it is not done by a data provider.

In MC holidays are seen in the astronomical sense: when you plot a chart all data corresponding to the holiday's date (exchange timezone) is ignored.

It is done this way because holidays start and end at different time: there are differences in different holidays on the same exchange, same holidays on different exchanges (or from different data providers). There is no single system.

We plan to add the possibility to set start and end time for holidays in the future. This should solve the problem.

If you use IB, IQFeed or TS as a data provider, then to get ES futures data after holiday starting at 17-00, you can just turn the holidays option off. (for ex. set Holday List as 'None' in the exchange settings in the QM).

Could you please explain what do you mean by : "The solution is obviously deleting all holidays in QM, but this creates another problem when mixing overnight data with the regular session data since dates will not match."?
Please send a screenshot with some explanations.

User avatar
Henrik
Posts: 140
Joined: 13 Apr 2010
Has thanked: 25 times
Been thanked: 11 times

Postby Henrik » 28 May 2010

Feature suggestion:
Please add "Seconds" to the "Bar magnifer" (see picture) for backtesting strategy.

Because recording Ticks (bid/ask/trade) from several instruments is a little bit unstable, but seconds are OK.
Attachments
secondadd.JPG
(56.38 KiB) Downloaded 17751 times

User avatar
Henrik
Posts: 140
Joined: 13 Apr 2010
Has thanked: 25 times
Been thanked: 11 times

Postby Henrik » 28 May 2010

Another Feature suggestion:

Add a new Field "Description" for paramaters.

For example in the code:

Code: Select all

Inputs:
PosiSize(1,"Size of Posi");
And now you can see the Description "Size of Posi" when you hold over the mousepointer when you format Signal>=Parameters.

OK, Posisize is a bad example, but I have sometimes very complex Parameters, and after some weeks I have to look in the code what I write as commentar (in power editor).

User avatar
arnie
Posts: 1594
Joined: 11 Feb 2009
Location: Portugal
Has thanked: 481 times
Been thanked: 514 times

Postby arnie » 28 May 2010

Hi Dave.
(for ex. set Holday List as 'None' in the exchange settings in the QM).
Apparently you have a QM version different from mine (MC 6 beta 3) :?
Where is the "None" option on the Holiday list?


When I said:
The solution is obviously deleting all holidays in QM
It is because there is no other way to turn off the Holidays option.

but this creates another problem when mixing overnight data with the regular session data since dates will not match.
What I mean by this is what you said previously regarding the option that a MC future version will have:
We plan to add the possibility to set start and end time for holidays in the future.
Taking ES for an example, ES trades between 1530 and 1515 (next day) EST.
Holidays run between midnights.
So if today were an holiday in the states, ES futures would start the new day at 1700 EST as usual.

MC though, would not show this data since it would treat this day as an holiday. Only after midnight MC would start to retrieve ES data since the holiday would be officially over and a new day would began.

The problem would then be the lack of 7 hours of trading that MC woudn't show for that day on ES chart, because those 7 hours would be included on the holiday session, a session, that due to QM Holiday option, would not be shown on MC.

If I'm not mistaken, Dennis promised a on/off button for the Holiday option in a future MC version.

Regards,
Fernando

User avatar
geizer
Posts: 375
Joined: 16 Jun 2008
Has thanked: 40 times
Been thanked: 38 times

Postby geizer » 28 May 2010

Agree with Arnie on adding the On/Off Holiday button so it can be accessible from MC toolbar.


Another issue in MC Beta 3:

Two title Bar icons are not aligned with others. Noticed this immediately after customizing the appearance of windows (OS version: Windows 7). I changed the following: Title Bar width, Caption Buttons size, and border padding. The image is attached.

Thank you,
Pavel
Attachments
Misalignment.PNG
(2.22 KiB) Downloaded 17786 times

User avatar
Henrik
Posts: 140
Joined: 13 Apr 2010
Has thanked: 25 times
Been thanked: 11 times

Postby Henrik » 01 Jun 2010

Suggestion:
Write own text in status-bar of a chart (or/and title bar).

Why?
I have several charts open with different funtions und indicators. Sometimes I have problems to find quickly the right chart because some charts are similar.

How?
At time you can change some things in the status line. But you can not write a text. maybe it can be implemented?
(Alternate: change titel of the title bar of a chart)

Thank you for think about it, TS team!

User avatar
Henrik
Posts: 140
Joined: 13 Apr 2010
Has thanked: 25 times
Been thanked: 11 times

BUG-Report (IntrabarOrder & Variables)

Postby Henrik » 14 Jun 2010

BUG-Report (IntrabarOrder & Variables):

Hi,
i have been working with Intra-Bar Order Generation and found a bug.
Settings: “Allow unlimited entries and exits per bar”

Code:

Code: Select all

print("Line 1 rlong: ",(rlong)," Marketposition: ",marketposition," date: ",date," ",time);
...
vars:
rlong(false),
rshort(false),
iBBO(0),
iBBU(0),
tradingtime(false),
BentryO(false),
BentryU(false),
FentryO(false),
FentryU(false),
RATR(0),
Trail(0);
...
print("Line 65 rlong: ",(rlong)," Marketposition: ",marketposition," date: ",date," ",time);

//Enter
if (tradingtime and FentryO and BentryO and not rlong) then
begin
buy next bar at market;
Trail=Close-RATR;
rlong=true;

end;

print("Line 74 rlong: ",(rlong)," Marketposition: ",marketposition," date: ",date," ",time);

print("Line 111 rlong: ",(rlong)," Marketposition: ",marketposition," date: ",date," ",time);

Line 111 is the end of the Code.
On a Intra-Bar Order Generation (tradingtime=true, FentryO=true, BentryO=true and not rlong)
it executes an long order position just as it should.
The variables “Trail” and “rlong” are receiving new values but only temporary. WHY?

They keep there new values until the end of the code. Starting the code over again they have there old values back but not the new ones.
This happens over and over again until the end of the Bar. If the entry is still valid the variables Trail and rlong keep there new values permanently.
BUT now this is the big fail... when the entry is not valid at the end of the bar the variables Trail and rlong have there old values back and not the new assigned ones BUT the MarketOrder is out.
The variables values should not change from last Line 111 to first Line 1 but it does


Log from the print Lines
Line 1 rlong: FALSE Marketposition: 0.00 date: 1100330.00 906.00
Line 65 rlong: FALSE Marketposition: 0.00 date: 1100330.00 906.00
Line 74 rlong: FALSE Marketposition: 0.00 date: 1100330.00 906.00
Line 111 rlong: FALSE Marketposition: 0.00 date: 1100330.00 906.00
Line 1 rlong: FALSE Marketposition: 0.00 date: 1100330.00 907.00
Line 65 rlong: FALSE Marketposition: 0.00 date: 1100330.00 907.00
Line 74 rlong: TRUE Marketposition: 0.00 date: 1100330.00 907.00
Line 111 rlong: TRUE Marketposition: 0.00 date: 1100330.00 907.00
Line 1 rlong: FALSE Marketposition: 1.00 date: 1100330.00 908.00
Line 65 rlong: FALSE Marketposition: 1.00 date: 1100330.00 908.00
Line 74 rlong: TRUE Marketposition: 1.00 date: 1100330.00 908.00
Line 111 rlong: TRUE Marketposition: 1.00 date: 1100330.00 908.00
Line 1 rlong: FALSE Marketposition: 1.00 date: 1100330.00 909.00
Line 65 rlong: FALSE Marketposition: 1.00 date: 1100330.00 909.00
Line 74 rlong: TRUE Marketposition: 1.00 date: 1100330.00 909.00
Line 111 rlong: TRUE Marketposition: 1.00 date: 1100330.00 909.00
Line 1 rlong: FALSE Marketposition: 1.00 date: 1100330.00 909.00
Line 65 rlong: FALSE Marketposition: 1.00 date: 1100330.00 909.00
Line 74 rlong: TRUE Marketposition: 1.00 date: 1100330.00 909.00
Line 111 rlong: TRUE Marketposition: 1.00 date: 1100330.00 909.00
Line 1 rlong: FALSE Marketposition: 1.00 date: 1100330.00 910.00
Line 65 rlong: FALSE Marketposition: 1.00 date: 1100330.00 910.00
Line 74 rlong: TRUE Marketposition: 1.00 date: 1100330.00 910.00
Line 111 rlong: TRUE Marketposition: 1.00 date: 1100330.00 910.00
Line 1 rlong: FALSE Marketposition: 1.00 date: 1100330.00 911.00
Line 65 rlong: FALSE Marketposition: 1.00 date: 1100330.00 911.00
Line 74 rlong: TRUE Marketposition: 1.00 date: 1100330.00 911.00
Line 111 rlong: TRUE Marketposition: 1.00 date: 1100330.00 911.00
Line 1 rlong: FALSE Marketposition: 1.00 date: 1100330.00 912.00
Line 65 rlong: FALSE Marketposition: 1.00 date: 1100330.00 912.00
Line 74 rlong: TRUE Marketposition: 1.00 date: 1100330.00 912.00
Line 111 rlong: TRUE Marketposition: 1.00 date: 1100330.00 912.00
Line 1 rlong: FALSE Marketposition: 1.00 date: 1100330.00 913.00
Line 65 rlong: FALSE Marketposition: 1.00 date: 1100330.00 913.00
[...]

sptrader
Posts: 742
Joined: 09 Apr 2010
Location: Texas
Has thanked: 483 times
Been thanked: 274 times
Contact:

MC (IB data) Continuous Contracts Bug ?

Postby sptrader » 14 Jun 2010

I just noticed that if I use continuous contracts and 50 days back(30 min chart), I get the ES June contract prices, However if I use 100 days back, I get the correct ES Sept contract prices- see attached.
Bug ??????
Attachments
Continuous Contracts Problem.jpg
(54.02 KiB) Downloaded 17745 times

User avatar
Henrik
Posts: 140
Joined: 13 Apr 2010
Has thanked: 25 times
Been thanked: 11 times

BUG-Report (IntrabarOrder & Variables):

Postby Henrik » 15 Jun 2010

I think my last posting is hard to understand (BUG-Report (IntrabarOrder & Variables).

Here a summary:

Intrabarorder activated, bar magifer activatet.

MC did not save parameters in a variable, when intrabaroder activated.
Only at close of a bar, but only if the condition true.

Please fix this, because its a big bug!

User avatar
arnie
Posts: 1594
Joined: 11 Feb 2009
Location: Portugal
Has thanked: 481 times
Been thanked: 514 times

Re: BUG-Report (IntrabarOrder & Variables):

Postby arnie » 15 Jun 2010

I think my last posting is hard to understand (BUG-Report (IntrabarOrder & Variables).

Here a summary:

Intrabarorder activated, bar magifer activatet.

MC did not save parameters in a variable, when intrabaroder activated.
Only at close of a bar, but only if the condition true.

Please fix this, because its a big bug!

Please report this through the Feedback/Bug Report option on MC.

:wink:

Chi-Yuan Shih
Posts: 14
Joined: 25 Aug 2007

Optimize failed

Postby Chi-Yuan Shih » 16 Jun 2010

As attached JPGs, MC6 beta3 failed to optimize.

If I move the mouse to Optimization Criteria tab and click, error message prompts as in OptimizeFailed.JPG.

When I try to modify the startvalue and endvalue for optimization.
The program halted and then crashed. [OptimizeHalt.JPG]

問題事件名稱: APPCRASH
應用程式名稱: MultiCharts.exe
應用程式版本: 5.0.3276.203
應用程式時間戳記: 4bd6a1d1
錯誤模組名稱: StackHash_090c
錯誤模組版本: 6.1.7600.16385
錯誤模組時間戳記: 4a5bdadb
例外狀況代碼: c0000374
例外狀況位移: 000c283b
作業系統版本: 6.1.7600.2.0.0.256.48
地區設定識別碼: 1028
其他資訊 1: 090c
其他資訊 2: 090c8f54f4ba4e79e36bdaf67354c176
其他資訊 3: cbe7
其他資訊 4: cbe70d4dbcf21dbb3ff391945fbeef4c

[OS: WIndows 7 Traditional Chinese]
Attachments
OptimizeHalt.JPG
(95.86 KiB) Downloaded 17739 times
OptimizeFailed.JPG
(49.44 KiB) Downloaded 17751 times

User avatar
Henrik
Posts: 140
Joined: 13 Apr 2010
Has thanked: 25 times
Been thanked: 11 times

Postby Henrik » 16 Jun 2010

I have there also problems with modify the startvalue and endvalue for optimization.
I can only click one month back or forward, than the calendar-button is closing.

But there is no crash like Chi-Yuan Shih.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Re: Optimize failed

Postby Tresor » 16 Jun 2010

When I try to modify the startvalue and endvalue for optimization.
The program halted and then crashed.
You should normally be able to change the optimizable inputs as per attached picture to arrive at number of simulations greater than 1.

I suggest to speak directly to TSS help desk and they can remotely diagnose the issue of MC's halting.
Attachments
Optimize non Failed.JPG
(50.24 KiB) Downloaded 17739 times

aknip
Posts: 3
Joined: 16 Jun 2010

Renko chart type display problem

Postby aknip » 17 Jun 2010

Hello,

I'm using MC 6.0 beta 3.

The Renko display problem for longer timeframes (backtesting) seems still to exist:
For every instrument I chose the last couple of days are displayed correctly (appro. 7-10 days), but going back more to the past the chart is empty - no bars shown at all.

This was already mentioned in an earlier post (March) and commented as "will be fixed", but in beta 3 the problem still exists:
http://forum.tssupport.com/viewtopic.ph ... torder=asc


If I adjust the selected timeframe, e.g. by shifting the displayed chart data 3 month backwards or so, again the last 7-10 days are shown, the rest is not displayed.

This is reproducable for all instruments and all data sources (ASCII or Zen), thus corrupted tick data can be excluded.

The same happens with 3LB chart type, thus it seems to be a bug for all non-time-based chart types?

Kind regards from Germany
aknip

User avatar
Dave Masalov
Posts: 1712
Joined: 16 Apr 2010
Has thanked: 51 times
Been thanked: 489 times

Postby Dave Masalov » 18 Jun 2010

Another Feature suggestion:

Add a new Field "Description" for paramaters.
Dear Henrik,

We will implement this feature in the future.

Apparently you have a QM version different from mine (MC 6 beta 3)
Where is the "None" option on the Holiday list?
Dear Fernando,

The "None" option is situated in QM --> Tools --> Exchanges & ECNs --> Edit --> Holiday List

If I'm not mistaken, Dennis promised a on/off button for the Holiday option in a future MC version.
Agree with Arnie on adding the On/Off Holiday button so it can be accessible from MC toolbar.
Dear Sirs,

This feature will be implemented in the future versions of MultiCharts.

Another issue in MC Beta 3:

Two title Bar icons are not aligned with others. Noticed this immediately after customizing the appearance of windows (OS version: Windows 7). I changed the following: Title Bar width, Caption Buttons size, and border padding. The image is attached.
Dear Pavel,

This problem should disappear after restarting MultiCharts.

Suggestion:
Write own text in status-bar of a chart (or/and title bar).
Dear Henrik,

We will consider your suggestions in the future development.

User avatar
Dave Masalov
Posts: 1712
Joined: 16 Apr 2010
Has thanked: 51 times
Been thanked: 489 times

Postby Dave Masalov » 18 Jun 2010

When I try to modify the startvalue and endvalue for optimization.
The program halted and then crashed. [OptimizeHalt.JPG]
The Renko display problem for longer timeframes (backtesting) seems still to exist:
For every instrument I chose the last couple of days are displayed correctly (appro. 7-10 days), but going back more to the past the chart is empty - no bars shown at all.

Dear Sirs,

This bugs should be fixed in the beta 4 version of MultiCharts 6.0

User avatar
Dave Masalov
Posts: 1712
Joined: 16 Apr 2010
Has thanked: 51 times
Been thanked: 489 times

Postby Dave Masalov » 18 Jun 2010

Hi,
i have been working with Intra-Bar Order Generation and found a bug.
Settings: “Allow unlimited entries and exits per bar”
Dear Henrik,

Please try to use the IntraBarPersist keyword.

It is used in variable and array declaration statements, before a variable or array name, to specify that the value of the variable or array elements are to be updated on every tick.
If IntraBarPersist is not specified, the value will be updated at the close of each bar.

Usage
Declaration:[IntraBarPersist]Name(InitialValue1)

Please read the Help file for more information.

User avatar
RobotMan
Posts: 375
Joined: 12 Jul 2006
Location: Los Altos, California, USA
Has thanked: 31 times
Been thanked: 13 times
Contact:

Postby RobotMan » 18 Jun 2010

Hi Dave,
Please try to use the IntraBarPersist keyword.
You realize that IntraBarPersist is unreliable at the present time (MC v6 b3) because of multithreading capability of MC and has promised to be fixed by Andrew.

bowlesj3
Posts: 2180
Joined: 21 Jul 2007
Has thanked: 227 times
Been thanked: 429 times

Postby bowlesj3 » 19 Jun 2010

Actually, according to my research the problem you mention Bob is one if missing ticks. I found the post with a revised subject name.
http://forum.tssupport.com/viewtopic.ph ... sing+ticks
If ticks are missing it will mess up your volume info for sure (kind of obvious). The big question then becomes, "If MC is made so it processes every tick in every study and the user has a lot of studies, can it get all these done before the next tick arrives from the exchange?". More importantly, "if it can not keep up how will the user get assistance such that they can decide if they should 1/ get a faster machine or 2/ eliminate some studies or 3/ make changes to select studies to reduce processing and speed them up and how will they know if it worked.".

Intrabarpersist is confusing but I am pretty sure it works as it is suppose to. I have a post that explains it.
http://forum.tssupport.com/viewtopic.php?t=6871
the help explains it only partially.

It just occured to me that is a very good way to explain intrabarpersist using code that represents what MC is doing. I will add a post to the above thread right now.

User avatar
RobotMan
Posts: 375
Joined: 12 Jul 2006
Location: Los Altos, California, USA
Has thanked: 31 times
Been thanked: 13 times
Contact:

Postby RobotMan » 19 Jun 2010

Hi John,

You have done more research than anyone else on the forum concerning this issue and I trust your conclusions. I don't know what the problem is. I only know what the results are; unreliable volume counts intrabar in any indicator research I try to do even though the charts themselves update reliably and all the volume is there in each bar on the chart.

I think I remember SUPER doing some tests in TS and IBP works fine. Andrew addressed the issue as TS is single threaded and MC is multiThreaded.
http://forum.tssupport.com/viewtopic.php?p=29535#29535
so that's why I said what I did.

bowlesj3
Posts: 2180
Joined: 21 Jul 2007
Has thanked: 227 times
Been thanked: 429 times

Postby bowlesj3 » 20 Jun 2010

Hi Bob,

I can explain what is happening to cause your volume bars to be giving the incorrect readings,

Price on a bar is fine because the high tick comes through and the low tick comes through and the open tick and the close tick too.

Volime bars exists because some of the ticks have come through and each of those ticks provide part of the volume for the volume bar.

However if even one tick is missing for the bar then the total volume for that bar will be incorrect (the volume bar will not be high enought or tall enough).

John.

P.S.
Now this is interesting. If MC it faced with situations where ticks come in just way to fast no matter how fast our computer is (and as a result of having too many studies that we feel are critical) and if we are traders like myself that do not use volume at all (YET at least), then if MC was smart enough to bypass ticks that do not provide a high/low/open/close, them maybe skipping ticks is a good sneaky way to get around the high volume fast market problem (but only during those times and under total control of the trader and fully reported to them too). For me, when markets start to show 10 second bars that are longer than my stop loss setting I stay out until things settle down a bit. If MC could report better in this area I would know if MC is having trouble before such long bar extremes. It would be very useful if MC could report and control in this area,

MC could easily know to skip executing studies for ticks that are not high/low/open/close. Very easy code involved (however like I say, it must be an option the trader must decide to switch on if MC reports it is having trouble keeping up with the high rate of tick arrival).

So an interesting question for you Bob is this. "have you never found an open/high/low/close that was incorrect?". I think the only way to research this properly would be to feed the same data into MC and TS then compare them (maybe super could modify my script to do this since I do not have time at the moment). IF MC and TS are the same here then the question becomes, "Is MC currently smart enough to properly skip ticks that are not open/high/low/close, but unwisely doing this without notifying the trader?"

It should be interesting how this all pans out when MC6.? is complete. I am not surprised that MC 6.? is taking a while to finalize. This is a big issue to say the least.

User avatar
RobotMan
Posts: 375
Joined: 12 Jul 2006
Location: Los Altos, California, USA
Has thanked: 31 times
Been thanked: 13 times
Contact:

Postby RobotMan » 20 Jun 2010

Hi John,

Maybe I do not understand. In my testing I see each bar receiving ALL the data. However, that same data is missing from being represented in an indicator that uses IntraBarPersist.
Ie: I can set up a volume bar chart and in subgraph 2 each histogram of "Ticks" shows a full bar of volume. If I run an indicator that counts volume using IBP in subgraph 3 at the same time, it does not see all the volume.
If I run a second or third IDENTICAL chart window at the same time, I get different results in subgraph 3 using the exact same data.

So my conclusion is that, internally MC sees *all* the upticks and downticks volume, but the indicators using IBP are unable to calculate them consistently. Hopefully this will all be moot shortly.

bowlesj3
Posts: 2180
Joined: 21 Jul 2007
Has thanked: 227 times
Been thanked: 429 times

Postby bowlesj3 » 20 Jun 2010

Hi Bob,

I went over your response a few times and I am pretty sure it is the missing ticks problem now that I remember the research.

The reserve word ticks is reproting correctly (meaning that by the end of the bar when "barstatus = 2" this reserve word has the correct number). What happens is there are times when the study is not executed for every tick coming in and yes you should be using intrabarpersist for your own totalling you set up and yes it is working properly but if your study is not executed even one time for a tick in the mid bar point your total will be lower. So that explains why your independant count is lower. Let me show an example. Lets say there are 10 ticks in this bar. I represent each tick below.


Code: Select all

ticks your count
1 1
2 2
3 3
(notice the gap)
5 4
6 5
7 6
8 7
9 8
10 9
So if the study does not get executed for tick #4 then your count can not be correct. Your count would be even worse if you removed intrabarpersist. You must have it in there to be correct. But in addition to this the study has to have been executed on tick #4 in order for your total to be correct. But executions of the study are getting missed for ticks that are coming in. Maybe my terminology needs to be changed to "Missing study executions (for specific ticks that exist according to the ticks reserve word)". In other words is it not the ticks that are missing but rather the study executions that are missing. I guess that is my fault. Glad you brought the issue up again.

So your statement "the data is there" is correct (tick #4 is there), but it is in fact not getting prcessed by the study because the study is not getting executed for tick #4 (and there is on magical way for your count to get incremented if that study is not executed).

User avatar
RobotMan
Posts: 375
Joined: 12 Jul 2006
Location: Los Altos, California, USA
Has thanked: 31 times
Been thanked: 13 times
Contact:

Postby RobotMan » 20 Jun 2010

Hi John,

I was just about to return your PM, but then I noticed this post and thought it would be better to reply here.

Here is the post (including my PL code) where I actually did several instances of the same indicator in the same chart and should have gotten the same results.
http://forum.tssupport.com/viewtopic.php?p=29388#29388

As I read over your post, I believe we are talking about the same thing only from different angles.
Single threading: It's like a catcher on home plate trying to catch a single ball thrown by the pitcher.
Multi threading: It's like a catcher on home plate trying to catch balls thrown by all the infielders.
Intrabarpersist drops a few balls. Am I close?
I am thinking that Andrew needs to tell his code to only use single threading through 1 CPU for an indicator whenever IBP is involved because using wait states or using a buffer probably won't work. (disclosure: I am not a programmer and Andrew is probably shaking his head at the last statement.)

---------------------
I looked up the post from an old thread where we were talking about IBP and I think it might be better to continue the conversation there for posterity's sake.

bowlesj3
Posts: 2180
Joined: 21 Jul 2007
Has thanked: 227 times
Been thanked: 429 times

Postby bowlesj3 » 21 Jun 2010

Hi Bob, I think the thread above on IntraBarPersist (the one I created
http://forum.tssupport.com/viewtopic.php?t=6871
) would be good to continue it. The last post of that thread has the intrabarpersist logic which I derived from all my tests on intrabarpersist. I will also review your code and see if I can figure out how the intrabarpersist logic might effect it. I will get back to this topic tonight after trading.

User avatar
Dave Masalov
Posts: 1712
Joined: 16 Apr 2010
Has thanked: 51 times
Been thanked: 489 times

Postby Dave Masalov » 21 Jun 2010

Feature suggestion:
Please add "Seconds" to the "Bar magnifer" (see picture) for backtesting strategy.

Because recording Ticks (bid/ask/trade) from several instruments is a little bit unstable, but seconds are OK.
Dear Henrik,

Could you please precise why recording ticks is unstable?

User avatar
Dave Masalov
Posts: 1712
Joined: 16 Apr 2010
Has thanked: 51 times
Been thanked: 489 times

Postby Dave Masalov » 22 Jun 2010

I just noticed that if I use continuous contracts and 50 days back(30 min chart), I get the ES June contract prices, However if I use 100 days back, I get the correct ES Sept contract prices- see attached.
Bug ??????
Dear sptrader,

We could not analyze the situation, because there is no Custom Futures formula on the screenshot (it is in the Descripion),

However, such situation is possible, if you made one or both of following actions at the time between creating 50 daysback and 100 daysback charts:

- changing the formula in the Custom Futures settings, i.e. 12VOC to 11VOC,

- adding symbols which participate in the CF chart plotting, i.e future U0 was added before plotting the second chart.

Did you do anything like that?
Are you able to reproduce the problem, if you do not change CF settings at the time between plotting two charts as well as do not add/delete symbols from database?
Please try to do Reload for this window - does the problem disappear?

If you are able to reproduce the problem, please come to our Live Chat so we can connect to your PC remotely and analyze it.

You can also give us the CF formula (SymbolCode from Edit Custom Futures dialogue) and tell what ES futures do you have in your QuoteManager database.

User avatar
Henrik
Posts: 140
Joined: 13 Apr 2010
Has thanked: 25 times
Been thanked: 11 times

Postby Henrik » 22 Jun 2010

Feature suggestion:
Please add "Seconds" to the "Bar magnifer" (see picture) for backtesting strategy.

Because recording Ticks (bid/ask/trade) from several instruments is a little bit unstable, but seconds are OK.
Dear Henrik,

Could you please precise why recording ticks is unstable?

Open too much charts to record tickdata from (for example) all FX-pairs - after a few days MC chrashed.
I think because of the 32Bit-Limitation.

User avatar
Dave Masalov
Posts: 1712
Joined: 16 Apr 2010
Has thanked: 51 times
Been thanked: 489 times

Postby Dave Masalov » 22 Jun 2010

Dear Henrik,

Recording ticks with "Skip ticks with identical prices" option enabled is very similar to collecting seconds, because the majority of ticks within a second are with identical prices. Only ticks with different prices are saved in the database.

Could you please describe your crash problem in details?

If you are able to reproduce the crash problem, please come to our Live Chat so we can connect to your PC remotely and analyze the problem.

User avatar
Henrik
Posts: 140
Joined: 13 Apr 2010
Has thanked: 25 times
Been thanked: 11 times

Postby Henrik » 22 Jun 2010



Could you please describe your crash problem in details?

If you are able to reproduce the crash problem, please come to our Live Chat so we can connect to your PC remotely and analyze the problem.
Thank you, Dave.
Now I use only Quotemanager to record Ticks and not MC with open charts.
Now I have no more crashes.

User avatar
geizer
Posts: 375
Joined: 16 Jun 2008
Has thanked: 40 times
Been thanked: 38 times

Postby geizer » 22 Jun 2010

Another issue in MC Beta 3:

Two title Bar icons are not aligned with others. Noticed this immediately after customizing the appearance of windows (OS version: Windows 7). I changed the following: Title Bar width, Caption Buttons size, and border padding. The image is attached.
Dear Pavel,

This problem should disappear after restarting MultiCharts.
Dave,
Thank you for response.
Restarting MultiCharts (as well as OS Windows) has no effect - buttons remain misaligned.
Another screen shot is attached for convenience.
This is a non-issue, however. Just a cosmetic effect.

--
Pavel
Attachments
Icons are not aligned.PNG
(2.42 KiB) Downloaded 17761 times

User avatar
Dave Masalov
Posts: 1712
Joined: 16 Apr 2010
Has thanked: 51 times
Been thanked: 489 times

Postby Dave Masalov » 24 Jun 2010

Restarting MultiCharts (as well as OS Windows) has no effect - buttons remain misaligned.
Another screen shot is attached for convenience.
This is a non-issue, however. Just a cosmetic effect.
Dear Pavel,

We have added this bug to our fix list.

User avatar
arnie
Posts: 1594
Joined: 11 Feb 2009
Location: Portugal
Has thanked: 481 times
Been thanked: 514 times

Postby arnie » 27 Jun 2010

Hi.

Can we expect improvements regarding PLE, namely, detached windows so we can expand different study formulas on different monitors for easier debug/comparison?


Regards,
Fernando

User avatar
Dave Masalov
Posts: 1712
Joined: 16 Apr 2010
Has thanked: 51 times
Been thanked: 489 times

Postby Dave Masalov » 28 Jun 2010

Dear Fernando,

I have forwarded your suggestion to the developers. It has been added to our feature request list.

bowlesj3
Posts: 2180
Joined: 21 Jul 2007
Has thanked: 227 times
Been thanked: 429 times

Postby bowlesj3 » 28 Jun 2010

Regarding that old issue “Studies not getting executed on every tick.” I wrote a new script which is attached to both 1/ retest this problem and 2/ prove that intrabarpersist does work.

Regarding #2 it has proven that there is in fact no problem with “intrabarpersist”. This was done by creating a Global variable execution count as backup to the main IntraBarPersist execution count and testing to be sure that they never go out of line.

Regarding #1 it appears that MC 6.0 beta 2 is running much better than MC was performing before. However if you run the attached script on the 1-minute bars (and you run it for 10 or so minutes) it does produce the message
“The previous tick did not cause execution.”. Meaning the two execution counts in the study have fallen behind the ticks count. You get a chimes.wav sound to let you know. Note: if you do too many fileappend writes fileappend will eventually abort. This is why I put a maximum input parameter into the script. This will abort the script if you get more than the maximum number of these. Anyone running this may want to set this maximum very high and just let the file append abort. I also included this maximum because there is no point in running the script all day. Also, I have noticed that at times MC will run a long time without creating this error. Much better.

Also it seems that MC is consistently executing the study twice for the first tick of the bar and the last tick of the bar (it never did this conistently in the past). This occurs when the message “Dup tick count” is written to the file. There is no alert for this since it would drive the trader crazy. I am curious why TSS would have MC doing this.


For anyone new to this issue I do not recommend going back to the old threads on this issue below. However for anyone who was involved before these may be of useful reference.

Bobs thread that started the investigation.
http://forum.tssupport.com/viewtopic.php?t=6307&start=0
The last thread I created.
http://forum.tssupport.com/viewtopic.ph ... =duplicate

Again the new test script is below. It is much simpler than the big one I created for comparing with TS (which super was using).
Attachments
A_TestTicks_SimpleWithGVCntAsWell.zip
(4.39 KiB) Downloaded 739 times

User avatar
RobotMan
Posts: 375
Joined: 12 Jul 2006
Location: Los Altos, California, USA
Has thanked: 31 times
Been thanked: 13 times
Contact:

Postby RobotMan » 28 Jun 2010

Hi John,

Just wanted to publicly thank you for all the work you have put into researching this vital and urgent issue.

User avatar
Dave Masalov
Posts: 1712
Joined: 16 Apr 2010
Has thanked: 51 times
Been thanked: 489 times

Postby Dave Masalov » 29 Jun 2010

Regarding that old issue “Studies not getting executed on every tick.”
Also it seems that MC is consistently executing the study twice for the first tick of the bar
Dear bowlesj3,

Thank you for testing these problems. These are known issues which will certainly be fixed. However, this will take a considerable amount of time as profound changes in the program architechture are required.

bowlesj3
Posts: 2180
Joined: 21 Jul 2007
Has thanked: 227 times
Been thanked: 429 times

Postby bowlesj3 » 29 Jun 2010

Hi Dave,

Yes, I am not surprised actually. However it seems as though it is a lot better now than the last time I did these tests (at least when it comes to missing an execution of the study when a tick comes in).

Thanks,
John

P.S. We have to thank Bob for his persistance. Had he not have kept complaining (and had he not have created that first script I used as a base), I never would have thought to test this. I am not sure why I decided to test the ticks output against his count. Natural programmers curiosity I think.

User avatar
arnie
Posts: 1594
Joined: 11 Feb 2009
Location: Portugal
Has thanked: 481 times
Been thanked: 514 times

Postby arnie » 30 Jun 2010

Hi guys.

I know that beta 4 was just released, and I'm installing it at this very moment, but I've just receive this error when I closed beta 3.

Just want to know if this is an known issue.

Thanks,
Fernando
Attachments
MC_error.JPG
(38.29 KiB) Downloaded 17754 times

User avatar
Dave Masalov
Posts: 1712
Joined: 16 Apr 2010
Has thanked: 51 times
Been thanked: 489 times

Postby Dave Masalov » 01 Jul 2010

Dear Fernando,

If you receive this error message in Beta 4, please come to our Live Chat so we can connect to your PC remotely and analyse the problem.


Return to “MultiCharts”