Playing with the VMAR`s open research

[walterw 10]

Very nice explanations here Unicorn and also very nice inputs as well...

 

What I finded very interesting is this two vma`s the cyan and red ones how they perform and all the nuances we can get from there for trading strategies..

 

I had been trying to find from our actual vma`s who would be similar and cant get there... I believe we need to program exactly Unicorn`s code to get there...

 

Thanks for this great interaction ¡¡ cheers Walter.

[PYenner 10]

Unicorn

 

Thank you.

Mathematicians think in concepts and speak in shorthand.

So while I sort of understand rsi(n), I find myself needing examples.

 

If I asked two people to walk around the world and one went around the equator and one went around the poles, they would both have done what I said but neither would know if it was what I intended, but didn't say.

 

Ditto with stochastic_adx.

But if abs(cmo) is very close to adx then maybe it should be tried in place of adx for vma lines?

 

Yes there are many variables and Bemac gets the Nobel Prize until we know any better.

Cheers

PYenner

[unicorn 10]

I believe we need to program exactly Unicorn`s code to get there...

 

Thanks for this great interaction ¡¡ cheers Walter.

 

You are welcome, Walter.

 

I understand that Sparrow has already programmed the vma_2, so that you can get both the signal and the rainbow basis on Ninja Trader.

 

Looking forward to your research.

 

regards.

Unicorn.

[Sparrow 10]

Hi all,

 

thanks for the explanation unicorn, although I didn't catch all of it.

Hope I can reach such a level of insight into indicators somewhen.

 

Cheers

[unicorn 10]

 

cmo(n) = 2*rsi(n) -0.5; re-scaling of rsi in the range (-0.5, 0.5)

 

correction:

cmo(n) = 2* ( rsi(n) -0.5 ); for raw rsi i.e. in the (0,1) range; re-scaling of rsi in the range (-0.5, 0.5) and multiplying by 2

or

cmo(n) = 2* ( rsi(n) -50 ); for rsi in the (0,100) range; re-scaling of rsi in the range (-50, 50) and multiplying by 2.

 

Unicorn.

[unicorn 10]

I find myself needing examples.

Ditto with stochastic_adx.

 

Of course, Bruce.

 

Go ahead and plot the cmo(n). Compare it to the rsi(n).

 

I trust you know the code for stochastics. In the stochastics computation, replace the close by the adx. Now you have stochastics_adx.

 

Have fun. Try the stochastic_rsi = stoch( rsi(rsi_periods), stoch_periods)

 

cheers.

Unicorn.

[PYenner 10]

I trust you know the code for stochastics..

Not yet but it looks like I am going to find out.

You are going to make me "indicator literate" if you keep this up.

In the stochastics computation, replace the close by the adx. Now you have stochastics_adx. Have fun. Try the stochastic_rsi = stoch( rsi(rsi_periods), stoch_periods)

Getting the picture, wheres a brick wall for me to bang my head on?

 

Looks like Snipers T3 CCI needs to be recoded since it looks to be needed.

That needs one more array than MT4 allows, so will have to find out how to get extra arrays for the first time and find out the hard way what the tricks and traps are with those. Scuse me :crap:

 

Midas is calling me, so I try OBV, turns out to be hopelessly primative.

Find a bar with H-L=39 and volume=25, impossible, seriously comforting, NOT.

Scuse me :crap:

5min OBV contradicts 1min OBV so price is leaking into those suckers.

Scuse me :crap:

 

You know how it goes...

[unicorn 10]

You are going to make me "indicator literate" if you keep this up.

 

Hello Bruce, Walter and Sparrow;

 

I think you will like this. Sparrow if you have the time, please code it in Ninja Trader.

I believe that it will help in understanding the vidya behavior, and in the selection of the appropriate parameters.

 

Have fun.

Unicorn.

vidya periods.afl.txt

[walterw 10]

Believe me I am not doing this to drive up my postings count

Unicorn just modded the indicator, so I am trying to catch up.

Parameter for the stochastics period added to the NT version.

 

 

One question here Sparrow... what setting should I put to your indi in order to get something similar to the cyan and red lines from Unicorn...

 

Technically speaking, is it the same formula from Unicorn`s cyan and red lines ?

 

Thanks for this contributions Sparrow... I am positive we will get to the point of having Unicorns exact same template on NT as well...

 

I read on NT forum some nice entuthiasm on vma`s ¡¡

 

thanks again, cheers Walter.

[walterw 10]

Midas is calling me

 

 

So Bruce... the Midas bug byte you ?... jejejej

 

we can open a new thread if you wish on this, so we dont get mixed... I got badly bitten by midas too jejej.... anyway vma`s do rock man ¡¡

 

In the next weeks I believe we will have a complete finished structural system arround vma`s and obviuosly each trader will have some nice alternatives to suit their tastes...

 

By the way, I smell some of our folks will also buy some bmw`s series 6 soon ¡¡ jejej I like the convertible one ¡¡ jejej... take care budy... cheers Walter.

[PYenner 10]

we can open a new thread if you wish on this, so we dont get mixed... I got badly bitten by midas too jejej....

Too soon for a volume thread and will probably have to test out forex volume on excel before there is any chance of saying if it is any good compared to real volume. MT4 has many hidden features, but unless one turns up that gives better access to the tick data, then it has to be studied in excel.

[PYenner 10]

Unicorn

 

Bear with me please on some questions on the difference between what is said and what is intended.

 

In lamely describing the third and final stage of the ADX as the "ADX_Bars" thing, I was referring to the output = input/(HHV-LLV) stage which you appear to descibe as Stochastic (input) = input/(HHV-LLV)... (including the way HHV and LLV are obtained)... and I expect you are correct.

 

Now where you describe Bemacs vma as stochastic_adx = stoch (adx (adx_periods), stoch_periods) you appear to have mistaken the stochastic within the ADX to be a secondary external and additional stochastic. Bemac used a straight copy and paste of the VT code for an ADX and changed only the final smoothing ema into a Vidya ama. He did not add a secondary stochastic.

 

So I would attempt to describe his vma as vma = viya(adx)

If you go into the input details, then ema_periods would apply to smoothing and ADX_Bars would apply to the internal stochastic, meaning ADX_Bars has the same meaning as stoch_periods.

 

As I read your code, you are applying a secondary stochastic right after the ADX's internal stochastic.

This would likely result in a stronger horizontal, but it also imposes a double time delay from having the ADX's ema smoothing between the two stochastics, then a similar delay from the final vidya ama.

 

Now we get to the part about what did Chande mean but not say.

You take stochastic_adx to mean an adx (complete with internal stochastic) followed by a secondary stochastic, I hesitate on that step.

 

So where Chande describes "abs(cmo) is very close to adx" is he meaning "like an adx but without needing an internal stochastic to get the result".

So he then refers to ordinary adx as stochastic_adx. Worse, there has been plenty of evolution in adx's, so ???

 

It seems to me that Bemacs vma = viya(adx) and again the vidya is internal to the ADX, meaning it replaces the normal output ema with a vidya ama and there are no additions external to the adx, its an adx with ama output smoothing. Some early vma versions used something like an ema of an ama but that is another story.

 

A result similar to using two stochastics can probably be obtained from vma6 with the "VI squared" option turned on, but again it makes for later laddering not just a stronger horizontal. I am uncertain if the vma1 version you are using has a straight ama output or one of the "ema of ama" outputs. It was a porting scramble back then too.

 

If I appear to be dragging my feet it is partly because I am uncertain whether things are going forward or round in circles.

Sorry to make questions..

Wheres that club again, :crap:

PYenner

[PYenner 10]

Walter

Do you know of any free forex platforms or charting packages that have proper tick charts? They would be the alternative to excel for Volume work.

[unicorn 10]

you appear to have mistaken the stochastic within the ADX to be a secondary external and additional stochastic.

 

As I read your code, you are applying a secondary stochastic right after the ADX's internal stochastic.

 

Hi Bruce;

 

The adx does not have an internal stochastic. You have probably forgotten how it is computed; you need to freshen up your knowledge.

This is the code for adx computation:

 

function adx_1( adx_periods)

{

PlusDM= IIf(High>Ref(High,-1) AND Low>=Ref(Low,-1), High-Ref(High,-1),

IIf(High>Ref(High,-1) AND Low<Ref(Low,-1) AND High-Ref(High,-1)>Ref(Low,-1)-Low, High-Ref(High,-1), 0));

DIPlus= 100 * Wilders(PlusDM,adx_periods) / ATR(adx_periods);

 

MinusDM= IIf(Low<Ref(Low,-1) AND High<=Ref(High,-1), Ref(Low,-1)-Low,

IIf(High>Ref(High,-1) AND Low<Ref(Low,-1) AND High-Ref(High,-1)<Ref(Low,-1)-Low, Ref(Low,-1)-Low, 0));

DIMinus = 100 * Wilders(MinusDM,adx_periods) / ATR(adx_periods);

 

Diff= abs(DIPlus - DIMinus);

DISum= DIPlus + DIMinus;

ADX_value = 100 * Wilders(Diff/DISum, adx_periods);

 

return ADX_value;

}

 

ATR(adx_periods) is the average true range, which in your mt4 code is different from the actual formula (of the TR true range), perhaps on purpose.

 

In order to understand what I am saying you should test this adx code (plot the result), alongside with the adx function of mt4. You need to know what your code is doing. Then plot the result of your computation as coded in your vmar code and compare it to the stochastic_adx, as implemented by the code I have used. Finally, it is a good idea to plot the coefficients that will be used in the vidya formula and compare the results. I have already posted my code for the coefficients.

 

It should be noted that the Wilders moving average " Wilders(quantity, adx_periods) " is in effect an exponential smoothing average

with equivalent_ema_periods = 2 * wilders_period -1;

Test it; Your test should be:

 

// test code

{

w_periods = Param("periods",9, 2, 55, 1);

w=Wilders(C, w_periods);

 

equivalent_ema_periods = 2*w_periods -1;

z= EMA(C, equivalent_ema_periods);

 

Plot(w,"Wilders", colorBlue, styleLine);

Plot(z,"ema", colorRed, styleLine);

}

 

do the math:

coeff_ema = 2/( ema_periods +1) = 2/( 2*w_periods -1 + 1) = 1/w_periods

 

Of course, the coefficients of the wilders_ma are 1/w_periods and (w_periods-1)/w_periods

 

You need to know what your code is doing to these coefficients, because you are actually modifying these computations in order to fine-tune vmar3.

 

cheers.

Unicorn.

[PYenner 10]

Thanks Unicorn and you are right, the stochastic is external to the ADX code.

So much for working from swiss cheese memory.

That also explains why there were two normalization divisions, seemed like overkill.

 

I have used a lazy version of ema weighting that differs slightly from standard "ema_periods" there will be a conversion for it but I have not bothered working it out. So thats one more complication.

 

A minor quibble regarding describing say Maji's Vidya as being based on price eg Maji = vidya(stochastic(price))

The input is actually delta price over delta time, aka momentum in pips/min for a 1min chart.

so Maji = vidya(stochastic(momentum)) might be dimensionally more sound.

 

But then the normalization makes it dimensionless, more like a momentum index or ratio unless that is a contradiction in terms, arghh.

 

Dude, there is only one answer here, you take over the MT4 programming and let me get a holiday, a long one, a very long one...

 

As far as the maths goes I would much rather be tracking the dimensional content, x% momentum, y% price etc so I know for real just what it is telling me. There seems little point in determining that it is 100% true to formula while the formula does not also tell you what the quantity is that is being displayed. Your car speedo is in miles/hr or ks/hr, what use would a speedo be if it mixed distance in with speed the same way a "momentum" indicator mixes price in with momentum. Something tells me that there should be a way to remove the price content that is inserted by smoothing and get you back closer to pure momentum. There is too much "hubble bubble" ignorance and not enough science in indicator signals.

 

Also in MT4 some programs repaint the most recent bar to correct for missed ticks. The ema function is one that has to be run in a repaint loop. Their internal functions are a mix of repainting and non repainting loops. Even the Tardis could not do a reliable job of comparing apples1 with apples2 to check for errors, MT4 does not have a reliable time basis, some programs repaint more than one bar. The comparison you suggest is still possible, constructive and sensible, full stop. Equally, until dimensional analysis accompanies it you can still not quantify and qualify what the indicator displays, so it might as well be called shoes sizes for elephants as momentum, it is neither, it is unknown and unscientific. The maths is not in contact with reality and that makes it more of a silly mind game than a tool, fools paradise.

 

Forgive the frustration.

I am currently running 5 momentum oscillators and 4 of them are a heap of you know what. I can't seem to face MT4 programming at the moment, possibly because it looks like it cant go where I need to go, possibly because so much is unreliable and undocumented that 80% of my time gets wasted by it, the reconstituted market feed makes a joke of fine tuning anything, I am seeing ticks being bundled and delayed for 30 secs or longer, during that time your indicator shows you no change.

There are signs that the frustration could be terminal.

The subroutines or diy function calls that you take for granted are not available within MT4 indicator programs, looks like you have to #define functions and do them on something like an include file. Of course its not documented, you have to find other proggies to copy and go search a forum for a day. The last week has been wasted in just this way, the programs are written in russian and there are no english versions of them.

It is possible I have had a gutfull of the MT4 runaround.

Been swollowing too much crud for too long with MT4.

Sorry all for the tantrum.

I'm taking some time out and it might be permanent as far as MT4 goes.

[unicorn 10]

I have used a lazy version of ema weighting that differs slightly from standard "ema_periods" there will be a conversion for it but I have not bothered working it out.

 

Bruce;

 

you are using wilders_ma; it is standard in the internal computation of adx; do you remember that Wilder created the adx?

In my post I was reminding this fact to you.

 

Take it easy.

Unicorn.

[unicorn 10]

The subroutines or diy function calls that you take for granted are not available within MT4 indicator programs,

I'm taking some time out and it might be permanent as far as MT4 goes.

 

Bruce;

 

You can try Amibroker; there is a free version to test that the programming suits your taste and requirements. It even downloads free historical data for your testing.

 

Or you can try Ninja Trader. For that you need to ask Sparrow.

 

Take care.

Unicorn.

[PYenner 10]

Bruce;

you are using wilders_ma; it is standard in the internal computation of adx; do you remember that Wilder created the adx?

In my post I was reminding this fact to you.

Take it easy.

Unicorn.

That was one of the reasons MT4 vma1 could not use Wilders or even the ema function, the ema would have ended up in the template file along with the fantail lines. Instead the ema's were done using up the indicator arrays that are limited to 8 in total. Use 6 emas and you can't do an oscillator unless you leave out the two Histo bars. It gets that sick dude. Its not software its knobbleware. There are ways around most of the knobbles but they are poorly documented or not documented at all and with even the most basic stuff there is this crazy treasure hunt you have to go on to maybe find out what they don't tell you. I don't have time for never ending runarounds. Again sorry for tuntrums.

[walterw 10]

Walter

Do you know of any free forex platforms or charting packages that have proper tick charts? They would be the alternative to excel for Volume work.

 

 

Hi Bruce... Ninjatrader would be the way... it has tick charts and its free, also forex data is free there... so far had tested quite a lot and has a very friendly charting functions and no connection issues (very important )...

 

I would love to get a chart on Ninjatrader similar to Unicorn`s on amibroker... hope we get there...

 

The edge of using Tick charts on forex is being able to get a very good precise timing... even better than the 1 min we actually have on mt4...

 

I dont know how their coding language differs to mt4... but if we could migrate there, it would be worth the effort eventually... cheers Walter.

[walterw 10]

Bruce;

 

You can try Amibroker; there is a free version to test that the programming suits your taste and requirements. It even downloads free historical data for your testing.

 

Or you can try Ninja Trader. For that you need to ask Sparrow.

 

Take care.

Unicorn.

 

Unicorn, what datafeed are you using for the currencies futures charts ? thanks Walter.

[unicorn 10]

Unicorn, what datafeed are you using for the currencies futures charts ? thanks Walter.

 

Good day Walter;

 

IQFeed

The CME package.

 

take care

Unicorn.

[Sparrow 10]

Hi all,

 

I took unicorn's suggestion and moved the Vidya into a seperate indicator, that way it is more reusable as well.

Walter: parameter's should be ADXPeriod:3, StochPeriod:5, EMAPeriod:3, MAPeriod:2

The MAPeriod is not configurable in unicorn's version, it's just for more final smoothing, but most likely the default value is fine.

 

I had a minor bug in the Vidya as well.

 

Here's also my first attempt of the sniper fx cci, i dunno if it is correct, hope someone can verify it. I used someone else's T3, dunno how well the results match up.

 

Cheers

FantailVMA1_v1.3.zip

FXST3CCI_0.8.zip

[walterw 10]

Good day Walter;

 

IQFeed

The CME package.

 

take care

Unicorn.

 

Thanks Unicorn...

[Blu-Ray 10]

Hi Bruce... Ninjatrader would be the way... it has tick charts and its free, also forex data is free there... so far had tested quite a lot and has a very friendly charting functions and no connection issues (very important )...

 

 

Walter

 

What data feed are you using to get free forex data into NT?

 

Cheers

 

Blu-Ray

[walterw 10]

Hi all,

 

I took unicorn's suggestion and moved the Vidya into a seperate indicator, that way it is more reusable as well.

Walter: parameter's should be ADXPeriod:3, StochPeriod:3, EMAPeriod:5, MAPeriod:2

The MAPeriod is not configurable in unicorn's version, it's just so more smoothing, but most likely the default value is fine.

 

I had a minor bug in the Vidya as well.

 

Here's also my first attempt of the sniper fx cci, i dunno if it is correct, hope someone can verify it. I used some else's T3, dunno how well the results match up.

 

Cheers

 

This is great Sparrow ¡¡ I am testing them right now...

 

Now let me ask again inputs :

 

Unicorns _Cyan line

_ Red line

 

would the rainbow be built with the ma periods ?

 

thanks for this great contributions ¡¡ cheers Walter.