Nonlagging Tools - page 47

[zilliq]

You never sleep ????

It's 1:00 AM in France...

Just to be sure too. At the end, the first sum we have nlmprices[r-k]

is it the sum of all price or all price with m as a momentum (can be 1, 2 or more as with a RMI ?

I say that because it strange with my code when I put length>6 the NonLagMa is no more on the candles but below as it is not usual with other MA (Tillson, SSmoother and so on)

Thanks

Zilliq

[zilliq]

For example a nonlagma with a lenght of 4: It's ok

But with a length of 9 in blue is far below at the difference for example of a Tillson MA in red who stay on the candles with a lengh of 6 and who is more smoother

It is logical but the NonlagMa seems to be more responsive but less smooth is it right ?

[Mladen Rakic]

zilliq:
For example a nonlagma with a lenght of 4: It's ok

But with a length of 9 in blue is far below at the difference for example of a Tillson MA in red who stay on the candles with a lengh of 6 and who is more smoother

It is logical but the NonlagMa seems to be more responsive but less smooth is it right ?

zilliq

Yes. As a rule of thumb, the more responsive some filter is (unless it is a fitting / recalculating filter) the less smooth it will be

[zilliq]

I think I have a problem with my code, because with the length of 1, the NonlagMa is not on the price at the difference of your code

I will seek where is the problem it seems to be on the alfa

See U

Zilliq

[Mladen Rakic]

zilliq:
I think I have a problem with my code, because with the length of 1, the NonlagMa is not on the price at the difference of your code

I will seek where is the problem it seems to be on the alfa

See U

Zilliq

Zillq

Length 1 should be equal to price itself (each average must return price itself for length 1)

[zilliq]

Hi Mladen,

I hope you're fine,

I think I succeed in coding the Nonlagma

And I see something very curious: Its seems to very similar to a T3 Tillson as you see on my graph (in purple the nonlagma and in blue a T3). Do you ever see that and do you have an explanation ? T3 is smoother and have the same "Nonlag" shift. I thought that Nonlagma will be better.

Perhaps a stupid question but what is for your the "best" MA (T3, Oma, NonlagMa, Hull etc...), who means the best ratio smooth/Nonlag ?

Have a nice night and thanks for your help

Zilliq

[Mladen Rakic]

zilliq:
Hi Mladen,

I hope you're fine,

I think I succeed in coding the Nonlagma

And I see something very curious: Its seems to very similar to a T3 Tillson as you see on my graph (in purple the nonlagma and in blue a T3). Do you ever see that and do you have an explanation ? T3 is smoother and have the same "Nonlag" shift. I thought that Nonlagma will be better.

Perhaps a stupid question but what is for your the "best" MA (T3, Oma, NonlagMa, Hull etc...), who means the best ratio smooth/Nonlag ?

Have a nice night and thanks for your help

Zilliq

zilliq

Every average, if you compare short periods looks like some another average too (that is why, for example, adaptive averages look like any other average when the period is short - they simply do not have the "room" to show what are they actually like)

Only when the calculation periods are long enough, then you can see the difference. Here is a comparison of a 25 period T3 and nonlag ma - the difference is going to grow and grow as the period grows - orange is non lag ma green is T3

[zilliq]

Thanks Mladen,

I understand, and it's very clear on your graph

It's strange because when I use the nonlagma I code on my graph it's not so close of the price as on your graph (blue is T3 purple is Nonlagma) here with a period of 20

I know it isn't the same code language as MT4 but can you confirm the calculation of the alfa that we apply on each price:

*************************************

pi = 3.14159265358979323846264338327950288

Cyclee=4

Coeff = 3*pi

Phase = Length-1

Len = Length*Cyclee + Phase

weight=0

sum=0

for i=0 to Len

if i<=Phase-1 then

t = 1.0*i/(Phase-1)

else

t = 1.0 + (i-Phase+1)*(2.0*Cyclee-1.0)/(Cyclee*Length-1.0)

endif

beta = Cos(pi*t)

g = 1.0/(Coeff*t+1)

if t <= 0.5 then

g = 1

endif

alfa = g * beta

sum=close*alfa

weight=alfa

next

**************************

Because it seems I have a big difference and I don't understand why

After that we add all alfa*price on a period of "Length"

Same thing on alfa, we add all alfa on Length period

****************************

sum1=summation[Length](sum)

weight1=summation[Length](weight)

nonlagma=sum1/weight1

****************************

But I have not the same Nonlagma ???

Do I miss something on the MT4 code ?

Thanks a lot for your help

Zilliq

[Mladen Rakic]

zilliq:
Thanks Mladen,

I understand, and it's very clear on your graph

It's strange because when I use the nonlagma I code on my graph it's not so close of the price as on your graph (blue is T3 purple is Nonlagma) here with a period of 20

I know it isn't the same code language as MT4 but can you confirm the calculation of the alfa that we apply on each price:

*************************************

pi = 3.14159265358979323846264338327950288

Cyclee=4

Coeff = 3*pi

Phase = Length-1

Len = Length*Cyclee + Phase

weight=0

sum=0

for i=0 to Len

if i<=Phase-1 then

t = 1.0*i/(Phase-1)

else

t = 1.0 + (i-Phase+1)*(2.0*Cyclee-1.0)/(Cyclee*Length-1.0)

endif

beta = Cos(pi*t)

g = 1.0/(Coeff*t+1)

if t <= 0.5 then

g = 1

endif

alfa = g * beta

sum=close*alfa

weight=alfa

next

**************************

Because it seems I have a big difference and I don't understand why

After that we add all alfa*price on a period of "Length"

Same thing on alfa, we add all alfa on Length period

****************************

sum1=summation[Length](sum)

weight1=summation[Length](weight)

nonlagma=sum1/weight1

****************************

But I have not the same Nonlagma ???

Do I miss something on the MT4 code ?

Thanks a lot for your help

Zilliq

How is cosine function working in your platform : is it using radians or degrees for an argument?

Metatrader's cosine is using radians

[zilliq]

We use radian too

May be an explanation: In the MT4 code we multiply alfa with price

I suppose this is the price, i périods before, no ?

And so we need to add, for example with a len of 5

alfa*close[4]+alfa*close[3]+alfa*close[2]+alfa*close[1] and alfa*close

or

alfa[4]*close[4]+alfa[3]*close[3]+alfa[2]*close[2]+alfa[1]*close[1] and alfa*close ?

And it's strange, because with a length of 1 the len is equal to 4 (4*1+0) and so the nonlagma can't be equal to close because we add the alfa*price of the last 4 periods

Thanks for your next comments

Zilliq

Is use your code, but to better understand I use the simpliest code of the NonlagMav3

double pi = 3.1415926535;

double Coeff = 3*pi;

int Phase = Length-1;

double Len = Length*Cycle + Phase;

if ( counted_bars > 0 ) limit=Bars-counted_bars;

if ( counted_bars < 0 ) return(0);

if ( counted_bars ==0 ) limit=Bars-Len-1;

for(shift=limit;shift>=0;shift--)

{

Weight=0; Sum=0; t=0;

for (i=0;i<=Len-1;i++)

{

g = 1.0/(Coeff*t+1);

if (t <= 0.5 ) g = 1;

beta = MathCos(pi*t);

alfa = g * beta;

price = iMA(NULL,0,1,0,MODE_SMA,Price,shift+i);

Sum += alfa*price;

Weight += alfa;

if ( t < 1 ) t += 1.0/(Phase-1);

else if ( t < Len-1 ) t += (2*Cycle-1)/(Cycle*Length-1);

}

if (Weight > 0) MABuffer[shift] = Sum/Weight;