A True Composite Symbol for Multicharts.

[synonym 10]

Hi there

 

I'm a real coding novice and wondered if any of you experts (or non-experts, i'm not choosey!) out there could help me.

 

I want to be able to create a custom symbol in order to chart the spread differential between two (or ideally 3 or more) symbols in Multicharts.

 

I would want to be able to weight the consistuent symbols, so for example have symbol 1 as 1.00, symbol 2 as -0.50 and symbol 3 as -0.50. I would want the function to chart the differential between those symbols as one symbol (and not as an indicator). This would then allow me to utilise all of the indicators and functionality in MC to analyse this true composite symbol.

 

I know i can do this myself, through merging symbol data in excel and using ASCII format to bring it in to MC, but this is messy and timeconsuming. Ideally the composite symbol would be able to import the data from whatever dataprovider i am using and picking the data up from the selected symbols via quotemanager.

 

Someone has created a composite symbol (CompoSymbol) indicator on the MC forum and this can chart the differential between two symbols, but it does this as an indicator and so means you cannot have other indicators/analysis based on it.

 

I was wondering whether,

1. Is this actually possible in MC?

2. Has anyone has already had the need for this and so already done the coding?

3. If so, would they be kind enough to share it?

 

Looking forward to see if anyone can help or advise.

Regards

Syn

[Tams 20]

...

Someone has created a composite symbol (CompoSymbol) indicator on the MC forum and this can chart the differential between two symbols, but it does this as an indicator and so means you cannot have other indicators/analysis based on it...

 

Syn

 

 

 

 

Andrew already given you the solution, but I don't think you understood the process.

 

 

Let me try again here, with step-by-step instruction:

 

 

Step One:

 

create a function named spread_d1d2 with the following code:

 

spread_d1d2 = c data1 - c data2;

 

 

Step Two:

 

In your chart (with the 2 data series), apply any indicator you want to use for your analysis.

 

In the "Format Study" window, enter spread_d1d2 as the Price.

 

I have attached the moving average study as illustration.

 

 

HTH

 

 

Edited July 12, 2009 by Tams

[BlowFish 10]

I wonder what the practical limit is of MC? In neoticker (which I no longer use) you could build your own indexes from the constituent components as the DAX dosent have a $TICK maybe I'll have a go at making one and seeing if MC can handle it.

 

Another thought does data2 .. data30 .. dataNN generate indicator updates on ticks and is there a limit on NN?

[synonym 10]

Andrew already given you the solution, but I don't think you understood the process.

 

 

Let me try again here, with step-by-step instruction:

 

Hi Tams,

that's fantastic. You're right Andrew did reply, but being a complete novice at coding i didn't realise the potential of what he was saying.

 

I really hope what you've mentioned works as i'd like it to. That'd be great! Don't take that as though i'm doubting you by the way. I'm just happy that you've probably solved my problem in such a simple way. I'll try it and let you know how i get on.

 

I assume that if i wanted to include more than two symbols in the spread, or make an equity spread where there is a partial ratio between the symbols, e.g. cdata1 - 0.70cdata2 - 0.30cdata2, i would simply include this in the code for the function. Am i right? I'm guessing i might be right in principle, but wrong in code! :haha:

 

Thanks again for your helpful reply.

Syn

Edited July 13, 2009 by synonym

[synonym 10]

I wonder what the practical limit is of MC? In neoticker (which I no longer use) you could build your own indexes from the constituent components as the DAX dosent have a $TICK maybe I'll have a go at making one and seeing if MC can handle it.

 

Another thought does data2 .. data30 .. dataNN generate indicator updates on ticks and is there a limit on NN?

 

It'd be interesting to know how you get on if you do try it. It's a really useful function, as not all data providers have the groupings/sectors, etc that you might want to analyse.

 

I can't see why dataNN would not update on ticks, if you are charting on ticks, but then again, i am far from the best qualified person to answer that question.

 

Syn

[Tams 20]

...

I assume that if i wanted to include more than two symbols in the spread, or make an equity spread where there is a partial ratio between the symbols, e.g. cdata1 - 0.70cdata2 - 0.30cdata2, i would simply include this in the code for the function. Am i right? I'm guessing i might be right in principle, but wrong in code! :haha:

 

Thanks again for your helpful reply.

Syn

 

 

 

YUP... you've got it.

 

It is as simple as that !!!

 

 

close data1 - 0.70 * close data2 - 0.30 * close data3

 

 

note: * is the multiplication sign.

 

 

 

 

 

p.s. use brackets to group items together for easier identification:

 

close data1 - ( 0.70 * close data2 ) - ( 0.30 * close data3)

 

 

.

Edited July 13, 2009 by Tams

[BlowFish 10]

Tams do you happen to know if a new print on data2 will generate an indicator update and do you happen to know how you check which data stream generated the update if it does?

[Tams 20]

Tams do you happen to know if a new print on data2 will generate an indicator update and do you happen to know how you check which data stream generated the update if it does?

 

 

MultiCharts technical support had posted an explanation at their site. (I will try to look for the post.)

Basically any new tick will generate a marker that something has changed, and MultiCharts will update all the dependent calculations and plot the chart accordingly.

[BlowFish 10]

That sounds adequate to construct your own index.

 

I wonder how scalable it is, at times MC dosent feel that fast and efficient simply receiving storing and displaying data. Maybe OK for building the DJ 30 but I imagine the SP 500 or the Russel 2000 would not be practical.

[synonym 10]

YUP... you've got it.

 

It is as simple as that !!!

 

 

close data1 - 0.70 * close data2 - 0.30 * close data3

 

 

note: * is the multiplication sign.

 

 

 

 

 

p.s. use brackets to group items together for easier identification:

 

close data1 - ( 0.70 * close data2 ) - ( 0.30 * close data3)

 

 

.

 

That's brilliant Tams, thank you! I've been unable to get the time to try this since your initial post. But i have today off from work and so i'll be doing it today.

Cheers

Syn

[synonym 10]

Hi Tams

i'm very please to say that i have got the spread function working perfectly along with various indicators based upon it. So, again thanks very much for your help!

 

The spread_d1d2, obviously plots as an indicator. So now the challenge is for me to find out whether i can base backtesting on this, using the spread_d1d2 indicator as the symbol price to use in the strategy, e.g. for the price of spread_d1d2 to be used as entry and exit points. I'm guessing that this may be rather difficult or impossible, because MC sees spread_d1d2 as an indicator and not a symbol and so MC might not actually allow you to treat it as a symbol.

 

Have you ever tried to use the spread_d1d2 indicator in such a way? It'd be really helpful to know as i have not done any backtesting in MC before and so have a bit of a learning curve to negotiate at the same time!

 

Cheers

Syn

[Tams 20]

That sounds adequate to construct your own index.

 

I wonder how scalable it is, at times MC dosent feel that fast and efficient simply receiving storing and displaying data. Maybe OK for building the DJ 30 but I imagine the SP 500 or the Russel 2000 would not be practical.

 

 

EasyLanguage was not designed to build a 2000 stock index.

 

As a matter of fact, EasyLanguage "was" not designed to do a lot of things we do now. It started life at a time when people were only making daily charts; intraday data were not yet readily available to the mass (pre-broadband days). You can see some of the keywords still have the EOD legacy to their design. e.g. Volume, Ticks.

 

In terms of speed, MultiCharts is one of the few programs that can utilize a multi-core CPU. I believe the limitation lies with the data feed, and not the program. (e.g. IB only allows 99 concurrent symbols for most users).

[Tams 20]

Hi Tams

i'm very please to say that i have got the spread function working perfectly along with various indicators based upon it. So, again thanks very much for your help!

 

The spread_d1d2, obviously plots as an indicator. So now the challenge is for me to find out whether i can base backtesting on this, using the spread_d1d2 indicator as the symbol price to use in the strategy, e.g. for the price of spread_d1d2 to be used as entry and exit points. I'm guessing that this may be rather difficult or impossible, because MC sees spread_d1d2 as an indicator and not a symbol and so MC might not actually allow you to treat it as a symbol.

 

Have you ever tried to use the spread_d1d2 indicator in such a way? It'd be really helpful to know as i have not done any backtesting in MC before and so have a bit of a learning curve to negotiate at the same time!

 

Cheers

Syn

 

 

Backtesting will give you varied results.

 

Bear in mind, backtesting is not REAL.

At least not "realistic" in 99% on the software on the market today.

 

The only way to make a "real" test is if the software can stream the data (with bid/ask) as if in real time.

 

 

If you base your backtesting on EOB and not IOG, you can obtain some workable results.

 

Enjoy!

[synonym 10]

Of course. I would only be wanting to do relatively simple backtesting. I like to keep things simple and i don't think complicated TA is useful on spreads. It's more about wanting to get a feel for what things are useful and how and what things are not.

 

Can you please explained what you mean by EOD or IOG?

 

And from your reply, i assume your implying that i can backtest on the basis of using the spread indicator in place of a symbol?

 

Cheers

Syn

[Tams 20]

EOB = End of Bar

EOD = End of Day

IOG = IntrabarOrderGeneration

[Tams 20]

...

And from your reply, i assume your implying that i can backtest on the basis of using the spread indicator in place of a symbol?

 

Cheers

Syn

 

 

see answer from prev post...

 

Backtesting will give you varied results.

 

...

 

If you base your backtesting on EOB and not IOG, you can obtain some workable results.

 

Enjoy!

 

 

p.s.

"Workable results" does not suggest "accurate/dependable/reliable/repeatable results".

With the current technology, backtesting cannot give you "accurate results".

Understanding the mechanics of "backtesting" can help you to "work" with the results you get.

Edited July 16, 2009 by Tams

[synonym 10]

Thanks for that Tams. You#'ve really helped me out with this thing.

Cheers

Syn

[BlowFish 10]

EasyLanguage was not designed to build a 2000 stock index.

 

As a matter of fact, EasyLanguage "was" not designed to do a lot of things we do now. It started life at a time when people were only making daily charts; intraday data were not yet readily available to the mass (pre-broadband days). You can see some of the keywords still have the EOD legacy to their design. e.g. Volume, Ticks.

 

In terms of speed, MultiCharts is one of the few programs that can utilize a multi-core CPU. I believe the limitation lies with the data feed, and not the program. (e.g. IB only allows 99 concurrent symbols for most users).

 

As an old supercharts user I appreciate it was very different in those days. You may be right, though MC certainly has odd bottlenecks (rasterising used to be very slow pre V5.0 not sure how it is now). Anyway I guess if you are going to do this sort of work Neoticker would be worth considering. The multi core stuff is mainly (only?) for optimising strategies I think? Some of the core tasks (receiving, storing and displaying data) still feel 'sluggish' to me.

 

Having said that it would be fun to do the DJ 30 and see if that works efficiently.

 

As an aside imho some of MC's biggest weaknesses are due to retaining old TS 'features' obviously EL is a massive plus point but some of the other architectural and UI design decisions where hampered by basing them on an ancient (for software) application.

[Tams 20]

As an old supercharts user I appreciate it was very different in those days. You may be right, though MC certainly has odd bottlenecks (rasterising used to be very slow pre V5.0 not sure how it is now). Anyway I guess if you are going to do this sort of work Neoticker would be worth considering. The multi core stuff is mainly (only?) for optimising strategies I think? Some of the core tasks (receiving, storing and displaying data) still feel 'sluggish' to me...

 

 

MultiCharts can use multi-core in backtesting/optimization as well as real time charting.

 

You can see a difference if you have multi-screen, lots of charts, and lots of indicators.

Each chart is assigned to a CPU core. If you have a quad-core CPU, you can see the workload is distributed over all 4 cores.

p.s. you can verify this by looking at the CPU Performance graphs in your Task Manager.

 

IO tasks are handled by one CPU. This is a PC hardware and OS logistics limitation, not a MultiCharts issue. This architecture is shared by most computers, except the mainframe and supercomputers.

 

 

.

Edited July 16, 2009 by Tams

[BlowFish 10]

MultiCharts can use multi-core in backtesting/optimization as well as real time charting.

 

You can see a difference if you have multi-screen, lots of charts, and lots of indicators.

Each chart is assigned to a CPU core. If you have a quad-core CPU, you can see the workload is distributed over all 4 cores.

However the IO tasks are handled by one CPU. This is a PC hardware and OS logistics limitation, not a MultiCharts issue. This architecture is shared by most computers, except the mainframe and supercomputers.

 

I didn't know that different charts used different threads,you live and learn. In the past 'IO' (receiving data, storing it and rasterising it) has been slugish (imo of course). Mind you the v5.0 patch notes mentioned significant improvements, 10 times for rasterising charts I believe. One way of looking at that is the old stuff was 'slow' well 10 times slower at least.

[synonym 10]

I wonder what the practical limit is of MC? In neoticker (which I no longer use) you could build your own indexes from the constituent components as the DAX dosent have a $TICK maybe I'll have a go at making one and seeing if MC can handle it.

 

Another thought does data2 .. data30 .. dataNN generate indicator updates on ticks and is there a limit on NN?

 

Hi BlowFish

just out of interest, does that mean neotciker can create such a custom symbol that can be used just as a real symbol, i.e. so you can make full use of neoticker's functionality on it?

Also, not being nosey, i just wondered why do you no longer use neoticker? Any particular weakness(es)?

Syn

[BlowFish 10]

Hi BlowFish

just out of interest, does that mean neotciker can create such a custom symbol that can be used just as a real symbol, i.e. so you can make full use of neoticker's functionality on it?

Also, not being nosey, i just wondered why do you no longer use neoticker? Any particular weakness(es)?

Syn

 

Neoticker is a great product though it has a pretty steep learning curve compared to good old easy language it seems pretty hard to me! It is fast and flexible but you pay for that power by having a lot to learn. The main reason I stopped using it (many years ago now) was I had some data management issues with odd holes appearing in my tick data. Looking back it probably wasn't that big of a deal though it was irritating. Take a look at TickQuest Inc. NeoTicker® - Professional Traders & Analysts NeoBreadth halfway down the page. Though I don't want to distract you from what you are doing!

[Tams 20]

NeoTicker is one powerful program; little known in the retail sector, it is mostly used by professionals and institutions. If I had not bought MultiCharts, it would be on the top of my list.

 

 

BTW, NeoTicker can do multi-stream tick playback. It is one of the few programs that can perform a forward backtest.

[synonym 10]

Neoticker is a great product though it has a pretty steep learning curve compared to good old easy language it seems pretty hard to me! It is fast and flexible but you pay for that power by having a lot to learn. The main reason I stopped using it (many years ago now) was I had some data management issues with odd holes appearing in my tick data. Looking back it probably wasn't that big of a deal though it was irritating. Take a look at TickQuest Inc. NeoTicker® - Professional Traders & Analysts NeoBreadth halfway down the page. Though I don't want to distract you from what you are doing!

 

I've just had a look. Looks absolutely perfect for my needs. But if it involves having to be more reliant on tougher coding than MC then i don't think it would be for me. But i'll definitely take a closer look.

 

Thanks BlowFish!

 

Syn

[synonym 10]

NeoTicker is one powerful program; little known in the retail sector, it is mostly used by professionals and institutions. If I had not bought MultiCharts, it would be on the top of my list.

 

 

BTW, NeoTicker can do multi-stream tick playback. It is one of the few programs that can perform a forward backtest.

 

I looked at Neoticker when i was buying MC. But i was put off by the complexity of it. In my perfect world, i could do the whole lot without having to touch any code. I guessed to use Neoticker you have to rely upon code more than you do using MC?