The perfect filter - page 5

[Алёша]

gunia:

..... in flat areas, increase the smoothing period, so there are not many wrong signals.

On the contrary, in "flat" areas we should decrease smoothing in order to decrease indicator lag, because "flat" is a lower timeframe saw, so filtering should be decreased.

[J.B]

gunia:

Good filter, I approve. The idea is concise and effective. Just shifting by momentum is a well-known topic, but on higher order increments, I haven't seen.

But anyway one should (adaptively) in flat areas, increase smoothing period there to avoid many wrong signals. That is, you will not be able to avoid it so easily:) Do ZZ by points where the colour changes and you will see how a simple TS would work with such a filter.

Thank you! Great idea!

I ask advanced gentlemen and ladies, to make corrections to the code, for I took the liberty of publishing it in kodobase))) I understand that this is very audacious, for a dummy like me, but he who does not risk does not drink champagne!

[Younga]

J.B:

Thank you! Great idea!!!

Please advanced gentlemen and ladies, make corrections to the code, as I took the liberty of publishing it in the kodobase))) I realise this is very audacious, for a dummy like me, but he who doesn't risk doesn't drink champagne!

the zigzag in the marked places does not coincide with the average line

[J.B]

Young:

the zigzag does not coincide with the average line in the marked places

These are the places where the indicator is wrong. It coincides with the average line, but the direction does not coincide. But as you can see statistically it is much more right than wrong, especially considering the simplicity of the idea. In general, it is a great idea to build a zigzag using the reversal signals! We should think of a way to build an integrator indicator based on such a zigzag to quantitatively see the growth and decline in equity in certain price points. To see where the filter lies and where it shows the truth. I am currently thinking of the best way to implement it.

[Younga]

J.B:
That's where the indicator is wrong. It coincides with the average line, but the direction is not the same. But as you can see statistically it is much more right than wrong, especially considering the simplicity of the idea. Generally speaking, it is a great idea to build a zigzag using the reversal signals! We should think of a way to build an integrator indicator based on such a zigzag to quantitatively see the growth and decline in equity in certain price points. To see where the filter lies and where it shows the truth. I am now thinking of the best way to implement it.

Now I've seen a zigzag on the clozes! (In trading, you don't have to wait for the candle to end - it's better not to use candles at all) anecdote (Candlelight Dinner - Romantic Hemorrhoid Cure)

See here http://forum.mql4.com/ru/45108/page43 AlexeyFX's post 13.01.2012 16: 26

[J.B]

Young:

Now I saw it - zigzagging across the clozes!

The picture shows the claws, but necessarily they are, I for example prefer HLC/3, kind of like MO Price in a minute)) In the turkey you can choose according to taste.

Look here http://forum.mql4.com/ru/45108/page43 AlexeyFX's post 13.01.2012 16: 26

I read this thread a couple of weeks ago. But my filter does not redraw! Neither does JJMA. Mine is weaker than JJMA because it's not adaptively adjusted, but nevertheless it is much more efficient than EMA. I understand the skepticism of ladies and gentlemen about such indices, everyone thinks that it automatically overdraws, but not this time. I wouldn't put a redrawer, it is not gentlemanly.

[Андрей]

J.B:

....... I realise that this is very audacious, for a dummy like me, but he who does not take risks does not drink champagne!

The kodobase is indeed bold. The idea is not bad, but we should add adaptivity and mark in ZZ not the trend direction, but whether it is correct, i.e. in those places where the indicator lies and where it tells the truth. Moreover, there may be some unspoken laws concerning the number of code lines necessary to avoid ridicule. I mean that self-respecting coder can be offended to download code less than for example 100 lines.

[J.B]

gunia:
In ZZ colour mark not the direction of the trend, but whether it is right, i.e. in those places where the indicator lies and where it tells the truth.

Maybe you are right, I will do it. But now I make a totalizer in a separate window for this indicator, to see how and where it works, it is a more quantitative way to mark where and how the reversal signals work.

Moreover, there may be some unspoken laws about the number of lines of code needed not to be made fun of. I.e. a self-respecting coder may be offended if you download code less than for example 100 lines.

I doubt it.

[J.B]

I have to admit that so far I lack knowledge how to fix my zigzag so that it would show the places where it lies and accordingly to make a competent adder, to evaluate the effectiveness of the filter((((

The output is just sum of all the additions in the zigzag and even that's not correct, in general I'm screwed for now.

If any of distinguished gentlemen and ladies, advise how to solve this problem, I will be grateful.

[Алёша]

Check out the Quick Test of Trading Ideas on the Chart, as far as I understand it is exactly what you need.