Indicator vs Signal

Questions about MultiCharts and user contributed studies.
janus
Posts: 746
Joined: 25 May 2009
Has thanked: 41 times
Been thanked: 82 times

Indicator vs Signal

Postby janus » 29 Jun 2010

Can someone please explain why plotting is not allowed in a signal study? I've been developing indicators as a function so I can call the same function in an indicator study for plotting, and a signal study for trading. On the face of it it appears silly to load two studies instead of one every time, especially if both utilise enormous amounts of resources, such as memory.

User avatar
Dave Masalov
Posts: 1712
Joined: 16 Apr 2010
Has thanked: 51 times
Been thanked: 485 times

Postby Dave Masalov » 30 Jun 2010

Dear janus,

Plotting is not allowed in a signal study, because of the difference in the calculations of signals and indicators. We understand that loading two studies instead of one can be very resource-demanding. This problem will be solved after the transition to 64 bits (most likely in 7.0 version of MultiCharts).

janus
Posts: 746
Joined: 25 May 2009
Has thanked: 41 times
Been thanked: 82 times

Postby janus » 30 Jun 2010

Thanks Dave. I look forward to the 64-bit version.

I don't understand why there should be a difference in the calculations. After all, don't we all expect them to be the same given we rely on the signal to behave the same way as the indicator? If there are differences in the calculations then it makes the requirement to plot in a signal even more relevant :!: Please clarify.

User avatar
Dave Masalov
Posts: 1712
Joined: 16 Apr 2010
Has thanked: 51 times
Been thanked: 485 times

Postby Dave Masalov » 30 Jun 2010

Dear janus,

Generally, signals and indicators are calculated the same way. However there are few nuances, such as the fact that there can be several signals in a strategy and, if all of them use plot1, it becomes not very clear, who will plot what.

Technically, the possibility of plotting can be added to signal studies. Yet it will require a lot of time and resources. We concentrate our development work on the transition to 64 bits, which will solve the memory issue when loading two studies instead of one.

User avatar
TJ
Posts: 6518
Joined: 29 Aug 2006
Location: Global Citizen
Has thanked: 957 times
Been thanked: 1883 times

Postby TJ » 30 Jun 2010

Does TS allow plot in strategy?
I haven't used TS in a long time, I do not remember if it was allowed.

User avatar
Dave Masalov
Posts: 1712
Joined: 16 Apr 2010
Has thanked: 51 times
Been thanked: 485 times

Postby Dave Masalov » 30 Jun 2010

Dear TJ

TS does not allow plot in strategy.

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

Postby Tresor » 30 Jun 2010

This problem will be solved after the transition to 64 bits (most likely in 7.0 version of MultiCharts).


I would like to apologize for being off-topic (I hope Janus forgives me), but when my eye catched ''64-bit'' I remembered something.

A few years ago I had access to data processing centre and tested a number of strategies on very-very-multi-CPU machines. At that time MC only worked on 32-bit OS, so I used Windows Server 32-bit as an operating system. The catch was a 32-bit OS could only work with 2 CPUs - a 32-bit OS did not see the remaining CPUs thus limiting MC's operations to only 2 CPUs.

If I can make a suggestion (before releasing an MC 64-bit version), please test it on a machine:
(i) with more than 2 true 64-bit CPUs inside like Itanium 9300 or Power6 (Xeons, i7s and AMD64s are quasi 64-bit)
(ii) with Windows Server 2008 R2 (there is only a 64-bit version of this OS).

If MC can work with more than 2 CPUs in full 64-bit hardware and software environment then MC acquires an advantage in corporate market (corporate guys like to use a full 64-bit architecture).

It could be a good selling point for MC (putting a screenshot of MC optimizing with a server with 4 / 6 / 8 / 10 / 12 CPUs with all cores working).

kiasom
Posts: 32
Joined: 30 Jun 2010
Has thanked: 49 times
Been thanked: 3 times

Re: Indicator vs Signal

Postby kiasom » 30 Jun 2010

Can someone please explain why plotting is not allowed in a signal study? I've been developing indicators as a function so I can call the same function in an indicator study for plotting, and a signal study for trading. On the face of it it appears silly to load two studies instead of one every time, especially if both utilise enormous amounts of resources, such as memory.


However, you can use the TL_new command for drawing trendlines. It covers some of the plotting ground:

value1 = TL_New(D[1],T[1],variableLastBar,D,T,variableThisBar);
TL_SetColor(value1,Magenta);
TL_SetStyle(value1,3);

janus
Posts: 746
Joined: 25 May 2009
Has thanked: 41 times
Been thanked: 82 times

Re: Indicator vs Signal

Postby janus » 30 Jun 2010

However, you can use the TL_new command for drawing trendlines. It covers some of the plotting ground:


Thanks kiasom but that's what I do sometimes already when I need to draw lines. This brings up another point. If we can draw trend lines in signal studies why not plot lines as well?

janus
Posts: 746
Joined: 25 May 2009
Has thanked: 41 times
Been thanked: 82 times

Postby janus » 30 Jun 2010

Generally, signals and indicators are calculated the same way. However there are few nuances, such as the fact that there can be several signals in a strategy and, if all of them use plot1, it becomes not very clear, who will plot what.


We don't have any issues running multiple indicator studies so why would there be a problem running multiple signal studies?

Technically, the possibility of plotting can be added to signal studies. Yet it will require a lot of time and resources. We concentrate our development work on the transition to 64 bits, which will solve the memory issue when loading two studies instead of one.


You just contradicted what was said by someone in response to another one of my concerns, namely TSS doesn't have enough resources. You just confirmed my suspicion, thank you. I repeat what I said before, I wish you had more resources to implement a lot of very useful and good ideas. Unfortunately, at the rate it's happening it will be many years if ever before many are done. That puts MC in a very vulnerable position as some competitors are moving much faster.

User avatar
Dave Masalov
Posts: 1712
Joined: 16 Apr 2010
Has thanked: 51 times
Been thanked: 485 times

Postby Dave Masalov » 01 Jul 2010

You just contradicted what was said by someone in response to another one of my concerns, namely TSS doesn't have enough resources. You just confirmed my suspicion, thank you. I repeat what I said before, I wish you had more resources to implement a lot of very useful and good ideas. Unfortunately, at the rate it's happening it will be many years if ever before many are done. That puts MC in a very vulnerable position as some competitors are moving much faster.


Dear janus,

We kindly ask you not to give your opinion as a fact. This is not the first time you do this. It is your pesonal opinion, that we do not have enough resources. We disagree. Here is our argument: please go to What's New section and take a look on how many features have been added and how many builds have been done the last 12 months. And then compare it with the competitors. Furthermore, we are working on several versions of MultiCharts at the same time.

Regarding the possibility of plotting in signal studies, we just do not share your point of view. We will not implement this feature. The problem with memory resources will be solved after the transition to 64 bits.

janus
Posts: 746
Joined: 25 May 2009
Has thanked: 41 times
Been thanked: 82 times

Postby janus » 01 Jul 2010

That's fine. That's all it was - an opinion, like all other statements I make.

I like to say this. I've been involved in the software development area all my working life. I have been involved in designing, developing, testing, installing, supporting and project management of a variety of complex applications across small to large enterprises and operating systems. I can very well appreciate the time and effort involved in something like MC.

As for not sharing my view that plotting in a signal study is necessary then that's a pity really. I'm almost certain one day you will come around to my point of view.

janus
Posts: 746
Joined: 25 May 2009
Has thanked: 41 times
Been thanked: 82 times

Postby janus » 01 Jul 2010

I just thought of a compromise. Why not alter the behavior of the compiler so that it displays a warning message if it contains commands appropriate only for indicators, ignores the command completely as if it wasn't there in the first place, and continues to compile successfully? That way we can just copy and paste indicator studies to signal studies, and add the appropriate code to place orders.

Another idea is to use pre-processor conditional directives like #ifdef to include or exclude statements in a study depending on whether it is a signal or an indicator study.

User avatar
TJ
Posts: 6518
Joined: 29 Aug 2006
Location: Global Citizen
Has thanked: 957 times
Been thanked: 1883 times

Postby TJ » 01 Jul 2010

That's fine. That's all it was - an opinion, like all other statements I make.

I like to say this. I've been involved in the software development area all my working life. I have been involved in designing, developing, testing, installing, supporting and project management of a variety of complex applications across small to large enterprises and operating systems. I can very well appreciate the time and effort involved in something like MC.
....


I am disappointed that you did not mention OpenVMS by name.


JK... ;-)>

User avatar
Dave Masalov
Posts: 1712
Joined: 16 Apr 2010
Has thanked: 51 times
Been thanked: 485 times

Postby Dave Masalov » 01 Jul 2010

I just thought of a compromise. Why not alter the behavior of the compiler so that it displays a warning message if it contains commands appropriate only for indicators, ignores the command completely as if it wasn't there in the first place, and continues to compile successfully? That way we can just copy and paste indicator studies to signal studies, and add the appropriate code to place orders.

Another idea is to use pre-processor conditional directives like #ifdef to include or exclude statements in a study depending on whether it is a signal or an indicator study.


Dear janus,

We understand your arguments. However, it would not be really a compromise as a lot of time is required to do this. We have more important objectives to concentrate on.


Return to “MultiCharts”