Weird RSI

[Russell]

I have taken a std RSI and coloured it over 75 and under 25 - when I checked it against an ordinary rsi the figures are all different. Here's the code - can anybody tell me why this should produce different figures?

[Russell]

Chart illustrating different figures


Anybody?

[JayF]

As seen here the RSIStandard.efs differs from the RSI under the Basic Studies menu. I'm investigating the difference, but you will also see that RSIEnhanced.efs has the same values.

Attached is an EFS that gets the RSI data a bit differently (and a bit more efficiently.) Instead of a Call to RSIStandard.efs, this creates an object called "study" and uses the getValue member to grab the RSI value.

if (study == null) study = new RSIStudy(nInputLength, "Close");
var vValue = study.getValue(RSIStudy.RSI);

[Russell]

RE: Reply to post 'Weird RSI'

Yes I can see there is a difference between enhanced and standard. Thank you


-----Original Message-----
From: [email protected] [mailto:[email protected]]
Sent: 19 March 2003 17:34
To: [email protected]
Subject: Reply to post 'Weird RSI'


Hello Russell,

~~~~~~~~~~~~~~~~~~~~~~~~~~~~

[Russell]

Lack of trigger alerts

I am not getting any trigger alerts from the attached. I think its set so that an ob/os true/false flag sets up when both rsi and stochastic goes over 65 or under 30. Its then released (triggered) when the macd crosses the zero line.
Does anyone have any ideas on this one.
Will attach chart in next comment

[Russell]

Chart of trigger less efs

[JayF]

I suspect the "&& !Strategy.isLong()" is the culprit. I suspect that Strategy.isLong() is true because of earlier signals being run through the formula. Are you using the eSignal Back Tester on this study? If not you could rem out all the Strategy. stuff. If you are using the Back Tester then you could do this...

Add the following to global variables section at the top...

var vLoading = true; 

Then change the following...

if (getBarState() == BARSTATE_ALLBARS) vLoading = false;

if (vLoading == false) {
    if (vOverSoldFlag == true && vHist > 0 && !Strategy.isLong()) {
        Strategy.doLong("Over Sold", Strategy.MARKET, Strategy.THISBAR);
        vOverSoldFlag = false;
        Alert.addToList(getSymbol(), "Go Long  "+close(), Color.black, Color.green);
        Alert.playSound("swoosh.wav");
        }

    if (vOverBoughtFlag == true && vHist < 0 && !Strategy.isShort()) {
        Strategy.doShort("Over Bought", Strategy.MARKET, Strategy.THISBAR);
        vOverBoughtFlag = false;
        Alert.addToList(getSymbol(), "Go Short  "+close(), Color.black, Color.red);
        Alert.playSound("train.wav");
        }   
} 

[JayF]

Looks like I had an error in my logic.

This line...

if (getBarState() == BARSTATE_ALLBARS) vLoading = false; 

Needs to be changed to...

if (getBarState() != BARSTATE_ALLBARS) vLoading = false; 

Below is the corrected efs

[Russell]

RE: Reply to post 'Weird RSI'

I would like to backtest the final version I get to so I will try this - and
thanks



-----Original Message-----
From: [email protected] [mailto:[email protected]]
Sent: 19 March 2003 20:52
To: [email protected]
Subject: Reply to post 'Weird RSI'


Hello Russell,

~~~~~~~~~~~~~~~~~~~~~~~~~~~~

[Russell]

No triggers or sounds still


Jay

Still nothing

[TURLIES]

Hi Russell,
I had the same problem months ago with the standardRSI creating different values than a normal RSI - all i did was use the enhancedRSI in all of my .efs studies.

If you create the study in the formula wizard using the RSI calculation there - it should produce the normal RSI figures.

Paul

[Russell]

RE: Reply to post 'Weird RSI'

Thanks - I assumed RSI was standard until Jay pointed it out. Obvious
really!

-----Original Message-----
From: [email protected] [mailto:[email protected]]
Sent: 20 March 2003 12:19
To: [email protected]
Subject: Reply to post 'Weird RSI'


Hello Russell,

~~~~~~~~~~~~~~~~~~~~~~~~~~~~

[Russell]

Format get time

How can I format the time in a trigger alert?
Currently I get the full date and time - I only want the time. I would also like the time output to be european time not american.

[Russell]

Missing triggers

Hi Jay

Image as discussed

[JayF]

Russell,

To review for the group, you had mentioned that you believed there were 5 missed triggers. Based on your screenshot, I don't believe this is the case. On the far left of the RSI in the screenshot, there are no RSI values outside the bands at the same time the Stoch is OB or OS in order to set those flags to watch for a MACD cross. I can't tell about the other arrow because the Triggered Alert list is covering it.

My best suggestion is to use the debugPrintLn() function to send values to the EFS Output Window. For instance...

if (vValue > 65 && vSLOW > 65) vOverBoughtFlag = true;
if (vValue < 30 && vSLOW < 30) vOverSoldFlag = true;

// debug code
debugPrintLn("The value of vOverBoughtFlag is " + vOverBoughtFlag);
debugPrintLn("The value of vOverSoldFlag is " + vOverSoldFlag); 

Just be sure to remove this debug code for your final version.

~Jay