Jump to content

Welcome to the new Traders Laboratory! Please bear with us as we finish the migration over the next few days. If you find any issues, want to leave feedback, get in touch with us, or offer suggestions please post to the Support forum here.

  • Welcome Guests

    Welcome. You are currently viewing the forum as a guest which does not give you access to all the great features at Traders Laboratory such as interacting with members, access to all forums, downloading attachments, and eligibility to win free giveaways. Registration is fast, simple and absolutely free. Create a FREE Traders Laboratory account here.

Tasuki

Recoding TS Indicators to Make Them More Efficient?

Recommended Posts

The gist of this post is that I'm looking for someone to help me re-code Tradestation indicators to make them more efficient. However, I don't even know if this is worth the time and effort. The idea for doing this began when I started a thread on the Tradestation forum which wound up turning into a discussion of the importance of re-coding TS indicators to make them more efficient. Here's the thread (you have to be a TS customer to get in, I think):

https://www.tradestation.com/Discussions/Topic.aspx?Topic_ID=81245

 

Anyway, I was wondering if anyone who knows easylanguage and Tradestation knows:

1) whether this is possible?

2) how you'd know if your new code was more efficient?

3) whether it would be worth the time and effort?

 

Here's a quote from the thread that gets to the heart of the matter:

 

-----------------------------------------------------------------------

"First thing is make sure your code is running efficiently. Clean up the code, make it efficient. And either use STO or something similar. STO reduces the # of calcs by as much as 90%. I have one indicator that I only update once per second. Another thing I do is to consolidate indicators/paintbars/everything running realtime into one indicator. You have to look at it from the perspective of reducing the # of instructions being sent to the cpu. Every if statement matters.

Further still, arrange your code so you don't keep repeating the same calc when nothing will change. For example, there is no need to keep repeating the lookback calcs over and over in stochastics. Do it once then modify the result if the current bar changes.

It is worth noting also that many "stock" TS indicators are built for convenience, not speed. I don't use userfunctions in realtime indicators. And sure as heck would never use one of those multi-ouput 3-4 layer of userfunction things. Make your own to do only what you need. Look at the TS function highest. Dig through all the layers to get to the real meat of it. In comparison, all you actually need is 4-5 lines of code. "

------------------------------------------------------------------------

 

So, my dear Traders Labmates, does this make sense? And is there anybody out there in TL-land that can do this (for a fee, of course)?

Share this post


Link to post
Share on other sites

You are unlikely to get more detailed help here than on the TS forum. The advices on the TS forum from the likes of solidus, mmillar, Javln, goose are all good and should be followed. The whole idea of optimization is to avoid calculations that you don't need or want and be sure you don't have to do it intrabar if you can help it. That is the basis of STO (same tick optimization).

 

The Auto Detect feature can be overwritten, and the optimization can be forced to be always on or off by selecting either the On or Off setting on the Advanced tab. The setting can also be set through an EasyLanguage attribute in the code. The syntax of the attribute is:

[sameTickOpt = Value] where Value can be either True or False

 

If you have EasyLanguage analysis techniques which do not require that calculations occur when multiple ticks are received at the same price level, then you might consider using this new optimization setting to reduce any unnecessary CPU load and improve overall TradeStation platform performance.

 

When plotting text, avoid updating the same text intrabar by:

You may want to plot Text only at the End of the Bar...

 

if (BarStatus(1)=2) then

begin

::::::

::::::

end;

Share this post


Link to post
Share on other sites

What is the study you want to speed up? The chances are that you will be able to, often significantly. Some of the inbuilt TS functions are pretty in efficient for a kick off. The big savings are if you can avoid loops and to shorten the stuff that's done on every tick.

Share this post


Link to post
Share on other sites

Blowfish, I 'm using mostly just the basic indicators, CCI, MACD, RSI, stochs, a few moving averages.

 

How do you know if an indicator is updating on every tick? I can't tell myself by looking at the code. Is there some way to tell the indicator only to update if the value changes, rather than every tick whether the value changes or not?

Share this post


Link to post
Share on other sites

How do you know if an indicator is updating on every tick? I can't tell myself by looking at the code. Is there some way to tell the indicator only to update if the value changes, rather than every tick whether the value changes or not?

 

Hi Taz

 

I've sent you an email about it, but I'll quickly explain it here for anyone else reading this.

 

Format the indicator and select the "Advanced tab", then select the "On" rather than "Auto Detect"..... but there is a bug within TS 8.3 build 1631 so this feaute won't work unless you get the fix via request from TS.

 

attachment.php?attachmentid=8310&stc=1&d=1223735066

 

Hope this helps

 

Blu-Ray

STO.png.553f06237230ebbf5bb29398ec6e3ecc.png

Share this post


Link to post
Share on other sites
Blowfish, I 'm using mostly just the basic indicators, CCI, MACD, RSI, stochs, a few moving averages.

 

How do you know if an indicator is updating on every tick? I can't tell myself by looking at the code. Is there some way to tell the indicator only to update if the value changes, rather than every tick whether the value changes or not?

 

If memory serves correctly (and things haven't changed) TS does not code moving averages efficiently.

 

For example the common way to do a 10 period simple moving average is in a loop adding each of the 10 closes and then dividing by 10.

 

The efficient way is to add 1/10th of the new close and subtract 1/10th of the close 10 bars ago.

 

There are many examples of this sort of optimisation that will yield significant speed improvements.

Share this post


Link to post
Share on other sites
If memory serves correctly (and things haven't changed) TS does not code moving averages efficiently.

 

For example the common way to do a 10 period simple moving average is in a loop adding each of the 10 closes and then dividing by 10.

 

The efficient way is to add 1/10th of the new close and subtract 1/10th of the close 10 bars ago.

 

There are many examples of this sort of optimisation that will yield significant speed improvements.

 

Blowfish, in what way is your approach more efficient?

Share this post


Link to post
Share on other sites
Hi Taz

 

I've sent you an email about it, but I'll quickly explain it here for anyone else reading this.

 

Format the indicator and select the "Advanced tab", then select the "On" rather than "Auto Detect"..... but there is a bug within TS 8.3 build 1631 so this feaute won't work unless you get the fix via request from TS.

 

attachment.php?attachmentid=8310&stc=1&d=1223735066

 

Hope this helps

 

Blu-Ray

 

Blue-Ray,

 

Just for clarification, are you suggesting that the "On" versus "Auto Detect" option should be used as the default setting only if the indicator has been optimized or regardless if it has been optimized or not?

 

Thanks

Share this post


Link to post
Share on other sites
Blowfish, in what way is your approach more efficient?

 

In my way there is an addition and a subtraction. The other way you have to maintain a loop (which means an extra variable that you decrement and compare to zero) (you do do loops from n-1 to zero rather than from 1 to n right?). You also have to do 10 summations within the loop and a division outside the loop. The saving is vast. For a 200 period MA my way will be at the very very least 600 times quicker!

 

Edit its not 'my' way its the....errr.....efficient way.

Share this post


Link to post
Share on other sites
In my way there is an addition and a subtraction. The other way you have to maintain a loop (which means an extra variable that you decrement and compare to zero) (you do do loops from n-1 to zero rather than from 1 to n right?). You also have to do 10 summations within the loop and a division outside the loop. The saving is vast. For a 200 period MA my way will be at the very very least 600 times quicker!

 

Edit its not 'my' way its the....errr.....efficient way.

 

Yipes, BF, that's amazing. Do you have your moving averages coded that way? If so, would you be willing to share? Or is this more efficient code available somewhere?

Many thanks,

Taz

Share this post


Link to post
Share on other sites
Yipes, BF, that's amazing. Do you have your moving averages coded that way? If so, would you be willing to share? Or is this more efficient code available somewhere?

Many thanks,

Taz

 

Errm actualy I am not sure I do tbh. Off the top of my head something like this should be close to working (I typed it straight into the message so may not be quite right)

 

inputs: period(20);

var: FMA(0);

 

if barnum > period then

begin

FMA= FMA + Close/period;

FMA= FMA - Close[period]/period;

end;

 

plot(FMA);

Share this post


Link to post
Share on other sites
Errm actualy I am not sure I do tbh. Off the top of my head something like this should be close to working (I typed it straight into the message so may not be quite right)

 

inputs: period(20);

var: FMA(0);

 

if barnum > period then

begin

FMA= FMA + Close/period;

FMA= FMA - Close[period]/period;

end;

 

plot(FMA);

 

BF, got an error message when I tried to verify. Highlighted "barnum" and said "Word not recognized by Easylanguage"

 

Is "barnum" a word in EL? I haven't seen it before but that doesn't mean much.

Thanks, Taz

Share this post


Link to post
Share on other sites

// compare this to TS's MA default 9 period
inputs: period(9);
var: 	MA_Sum	( Summation( Close, period ) ) , FMA(0); 

{  BF's code
if {barnum} currentbar > period then
begin
FMA= FMA + Close/period;
FMA= FMA - Close[period]/period;
end;
}

MA_Sum = MA_Sum + Close - Close[period] ; 
FMA = MA_Sum / period ;

plot1(FMA);

 

extensive testing have been done: https://www.tradestation.com/Discussions/Topic.aspx?Topic_ID=74179

Share this post


Link to post
Share on other sites

thrunner,

Thanks. so this is a simple moving average, obviously. I presume the same principles can be applied to exponential and weighted, right? I looked at the EL codes on TS for the simple and exponential MAs but I couldn't figure out, well, pretty much anything. Your code is so different from either of them that I can't figure out how to transpose the more efficient code to the exponential MA. Is it true that the exp. MA calculation is alot more complex (it's got exponents in it?)?

Share this post


Link to post
Share on other sites
thrunner,

Thanks. so this is a simple moving average, obviously. I presume the same principles can be applied to exponential and weighted, right? I looked at the EL codes on TS for the simple and exponential MAs but I couldn't figure out, well, pretty much anything. Your code is so different from either of them that I can't figure out how to transpose the more efficient code to the exponential MA. Is it true that the exp. MA calculation is alot more complex (it's got exponents in it?)?

 

Taz

 

The code thrunner has posted is the same one as I've sent you via email, as the boys on the TS forums have done extensive testing via different propositions and found this one to be the most efficient.

 

The code I sent you for the EMA is also found to be the most efficient by the TS guru's, I'll post it here so anybody else interested can see the code.

 

{XAverage-Fast Indicator}

 

[sameTickOpt = true]

 

Inputs: Price(Close),

Length(8);

 

Vars: SmoothingFactor( 2 / ( Length + 1 ) ),

XAverageFAST( Price ) ;

 

XAverageFAST = XAverageFAST[1] + SmoothingFactor * ( Price - XAverageFAST[1] );

 

Plot1(XAverageFAST,"Ema");

 

 

Hope this helps

 

Blu-Ray

Share this post


Link to post
Share on other sites

You might want to have a look here

 

http://en.wikipedia.org/wiki/Online_algorithm

 

I am not aware of any references or I would have offered them. My own knowledge was picked up many many many years ago when I worked in R&D as a software engineer. Most people that where there where far far smarter than me. Over time I was exposed to some clever stuff, lucky to work with a bunch of geniuses.

 

Every now and then I like to diddle with indicators and stuff to keep my hand in. I was rather pleased with myself (actually delighted) in coding Jerry Perls stuff without any loops. Normally as total volume changes you would re weigh everything with this new volume. It took a while but was quite satisfying, the VWAP was straight forward the standard deviations tricky, 3rd and 4th order moments so far beyond me. Luckily they where not part of his original work. There are algorithms for variance in the public domain but adding the volume weighted stuff was tricky and need the help of a mathematician.

 

I'm kind of rambling a bit here but it's an interesting topic.

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use.