10.4 MTF TMA SLOPE RSI Indicator

[mjws00]

Heya,

I find Bob's 10.4 RSI template amazing for signals and alerts. It has less false signals than the simple stoch equivalent. I'll get it nailed, but don't have the time and indicator development background some of you wizards have. I want something inclusive we can use for dashboards and EA's later.

Here is the basics I've started to put together. It has a long to ways go. It will have all the alerts, as well as an option to check a lower TF for a signal so you can auto jump down to H1 to see if RSI has already shifted there.

This is pre-alpha. i.e it doesn't work yet. But close. My timezone and schedule requires me to sleep. My gut feeling is I won't be the fastest to get this done, but it would help the community.

Anyone willing to help feel free to take the code and save a bunch of typing. Or discard if you're almost done your own version.

Regards,

Mike

EDIT: Alpha updated and now working very nicely. Now on to the bells and whistles.

[mjws00]

This now works nicely and first post updated. Bells and whistles to follow.

[zimezoom]

Thanks so much!

[mjws00]

Here is a more visual version. Based on Paul's TMA Slope MTF. It shows the interactions of the different RSI's. My intent was to view 1hr rsi movement as a leading indicator of the 4hr.

Update: Current bar glitch found and cured by CandleTiger. Works great. Thanks.

Be careful cranking up the bars, we have to get a slope for each bar, for each time frame, then calc rsi on the arrays, so heads up. It is instant with 500 bars on my box.

Enjoy.

[wojtek_amm]

Thanks so much! Great work

[spotdespot]

Here is a more visual version. Based on Paul's TMA Slope MTF. It shows the interactions of the different RSI's. My intent was to view 1hr rsi movement as a leading indicator of the 4hr. But all the extra flexibility to view everything every which way seems to mess with the last bar a bit. Sometimes its perfect, sometimes it gets clobbered by something (higher tf?)

Maybe someone who has coded a bunch of these super duper timeframe indi's can see the issue easily. My goal is more geared towards signals and logic not pretty pictures, so finding that glitch is secondary for me.

I'd be honored if someone spotted and fixed it. Or use it as is, and it will be fixed when time allows.

Maybe someone finds it useful. Even better an indi wizard gets a touch of inspiration.

Be careful cranking up the bars, we have to get a slope for each bar, for each time frame, then calc rsi on the arrays, so heads up. It is instant with 500 bars on my box.

Enjoy.

I wish I could help as it looked really good until I spotted the issue you were talking about. Having H1 RSI and H4 RSI heading in the same direction makes for a pretty healthy bit of confirmation. Unfortunately I am pretty clueless with indicator code.

[candletiger]

But all the extra flexibility to view everything every which way seems to mess with the last bar a bit. Sometimes its perfect, sometimes it gets clobbered by something (higher tf?)
.

What means clobbered? Do you have a screenshot? That would help to nail down the problem...

[mjws00]

Clobbered means messed up. The last bar gets corrupted somehow. It is correct and then gets over written wrong.

There is an error somewhere in the way the (0) bar is referenced. All that iBarshift() goodness is screwing something up. I didn't dig deep enough into it to fix it yet, or even really analyze it. I wanted something I could plug into an EA. Version A accomplished that. This was just intended as icing and a start on something pretty if people thought it might be useful.

Thanks for looking.

Mike

[candletiger]

Clobbered means messed up. The last bar gets corrupted somehow. It is correct and then gets over written wrong.

ok, got it. Easy to see once I put it on a chart.

you have 2 coding flaws:

Code: Select all

int start()
{
 .........  
    
   int counted_bars = IndicatorCounted();

   if ( counted_bars < 0 ) return ( -1 );
   if ( counted_bars > 0 ) counted_bars--;
   limit = Bars - counted_bars;
   if ( maxBars > 0 && limit > maxBars ) limit = maxBars;

  ArrayResize(TMA_Slope_Values,limit);

You resize your array to limit. limit is influenced by IndicatorCounted(). During runtime limit goes down from maxBars to usually 2. Do a trace, it's an eye-opener.
Solution: just move the ArrayResize line to init().

Code: Select all

double GetSlopeRSI(string symbol, int tf, int shift)
{
.........
   
   curRSI = iRSIOnArray(TMA_Slope_Values,limit,RSIPeriod,shift);

Same problem.
Solution: replace limit by ArraySize(TMA_Slope_Values).

Let me know if it solves your problem...

/ct

[mjws00]

CT, you are a king. Thank you. Will test, but I am sure you are right. That would throw a zero from iRSIOnArray(). Another trick to be aware of. I'm sure I've missed more in there as that was my first foray into the mtf onarray() craziness.