mladen: Well, first version of LSMA trend was posted long time ago ( this post : https://www.mql5.com/en/forum/180514/page34 ) and it was made only to show what some other indicator was about. In the mean time it was renamed (surprise, surprise ... ) and posted as something different while nothing at all was changed in it.

But not posting about that now .

The main problem (in my opinion) with it was the "oversensitivity" since all it is looking for is a slope of linear regression value (LSMA == linear regression value). This version is a possible way to avoid that "oversensitivity" and ads a kind of a filter on it that might help in avoiding "insignificant" changes.

Hi mladen,

Please elucidate more about this indicator. Does it start with nonLagMA? Is that how it originated? Then channeled with the slope value?

Then a 3rd component, a filter you say. Please elucidate more on how many components (3?) And what they do.

Whenever linear regression value slopes up value is added to "cumulative trend" value. When linear regression value slopes down that same value is subtracted from "cumulative trend" value. If the change of direction is within the channel width it is treated as an incomplete and undefined yet, only a previously valid trend direction is noted within the channel. If the value breaks out of the channel only then it becomes a "trend".

Hope this helps

Batchboy: Hi mladen,

Please elucidate more about this indicator. Does it start with nonLagMA? Is that how it originated? Then channeled with the slope value?

Then a 3rd component, a filter you say. Please elucidate more on how many components (3?) And what they do.

mladen: It is a linear regression value "in disguise"

Whenever linear regression value slopes up value is added to "cumulative trend" value. When linear regression value slopes down that same value is subtracted from "cumulative trend" value. If the change of direction is within the channel width it is treated as an incomplete and undefined yet, only a previously valid trend direction is noted within the channel. If the value breaks out of the channel only then it becomes a "trend".

Hope this helps

Thx :-)

You also mentioned you added a filter, can you tell me a little about?

BTW, 2nd parameter is "Channel Length", but it seems to adjust the width. Minor typo or I'm missing info?

You are right about the typo, but at the time of making it called it that way ...

As of filter : idea is simple - do not react immediately but wait some bars form trend change confirmation. If it continues int the same direction that it is a "new trend". Maybe the best is to look at the linear regression value colored on slope change and see what periods are tried to be avoided. Attaching one version that will show you a colored linear regression value on chart (the version that does not repaint) so you can compare it to the trend indicator and get a much clearer picture of what exactly is it all about

___________________________

PS: in the calculation I am using vinins shorter version of calculating linear regression value. He proved some 2 years ago that his way and the original way are mathematically identical and I like his way for the simplicity of it

Batchboy: Thx :-)

You also mentioned you added a filter, can you tell me a little about?

BTW, 2nd parameter is "Channel Length", but it seems to adjust the width. Minor typo or I'm missing info?

mladen: It is a linear regression value "in disguise"

Whenever linear regression value slopes up value is added to "cumulative trend" value. When linear regression value slopes down that same value is subtracted from "cumulative trend" value. If the change of direction is within the channel width it is treated as an incomplete and undefined yet, only a previously valid trend direction is noted within the channel. If the value breaks out of the channel only then it becomes a "trend".

Hope this helps

Thx mladen,

You say wait number of bars, how have you set it to determine that, fixed? And maybe that should be external variable?

I'm finding 1min TF and lengthy channel (1000) with so-so narrow width like 35 to have excellent character, good trend follow with little indecision. So even if you did give variable on waiting not sure anything could be better.

Main question I have now is, how does it not take srious the medium whiplashes, the opposite accumulation doesn't reverse the channel yet? Yet when subtle moment arrives to look the other way, how is this possible

Old saying, if something seems too good to be true, it probably isn't true...... But I've not been dis-enamored yet,so its eating at me, what am I missing as far as the drawbacks?

Number of bars is already a parameter (the is the ChannelLength - as we said it should say width but let us leave it as is for now, the length is referring to the way how it is calculated internally)

Please, play a bit with the indicator parameters to explore how it works. Also compare it to the one posted a few posts ago (the linear regression value) in order to familiarize how it works. I am afraid that nobody can answer all the questions of some indicators behavior in some situations (otherwise its name would be the "holly grail") Experiment and use what indicators are best at : immediate visual result of some settings

regards

Mladen

Batchboy: Thx mladen,

You say wait number of bars, how have you set it to determine that, fixed? And maybe that should be external variable?

I'm finding 1min TF and lengthy channel (1000) with so-so narrow width like 35 to have excellent character, good trend follow with little indecision. So even if you did give variable on waiting not sure anything could be better.

Main question I have now is, how does it not take srious the medium whiplashes, the opposite accumulation doesn't reverse the channel yet? Yet when subtle moment arrives to look the other way, how is this possible

Old saying, if something seems too good to be true, it probably isn't true...... But I've not been dis-enamored yet,so its eating at me, what am I missing as far as the drawbacks?

mladen: Number of bars is already a parameter (the is the ChannelLength - as we said it should say width but let us leave it as is for now, the length is referring to the way how it is calculated internally)

Please, play a bit with the indicator parameters to explore how it works. Also compare it to the one posted a few posts ago (the linear regression value) in order to familiarize how it works. I am afraid that nobody can answer all the questions of some indicators behavior in some situations (otherwise its name would be the "holly grail") Experiment and use what indicators are best at : immediate visual result of some settings

regards

Mladen

LOL, true!

I have been playing with. As far as experimental play I'm trying to get paid mql programmer to do a spec sheet I wrote for an EA that in optimization runs will do the ultimate "experimental play".

You are right. The indicator you posted does not repaint.

Here is the same thing but written differently (simpler code, no restriction in bars (it works for the whole chart now)) I think it will be clearer what does it do (from the code) now and it is more suitable for further working on now.

PS: kept all the logic same as in the original

regards

Mladen

Hey mladen - this is pretty good if I set risk to 1. Could you or someone else filter most of these arrows away via changeable semaphor range/period ?

voltagetoe: Hey mladen - this is pretty good if I set risk to 1. Could you or someone else filter most of these arrows away via changeable semaphor range/period ?

Hi Voltagetoe,

Small change just made the x1 and x2 user controlled, the x1 is overbought for normalized Wpr and x2 is oversold for normalized Wpr. The indicator is already using range to determine arrow distance from price.

mladen:Well, first version of LSMA trend was posted long time ago ( this post : https://www.mql5.com/en/forum/180514/page34 ) and it was made only to show what some other indicator was about. In the mean time it was renamed (surprise, surprise ... ) and posted as something different while nothing at all was changed in it.

But not posting about that now .

The main problem (in my opinion) with it was the "oversensitivity" since all it is looking for is a slope of linear regression value (LSMA == linear regression value). This version is a possible way to avoid that "oversensitivity" and ads a kind of a filter on it that might help in avoiding "insignificant" changes.

Hi mladen,

Please elucidate more about this indicator. Does it start with nonLagMA? Is that how it originated? Then channeled with the slope value?

Then a 3rd component, a filter you say. Please elucidate more on how many components (3?) And what they do.

Thanks.

...

It is a linear regression value "in disguise"

Whenever linear regression value slopes up value is added to "cumulative trend" value. When linear regression value slopes down that same value is subtracted from "cumulative trend" value. If the change of direction is within the channel width it is treated as an incomplete and undefined yet, only a previously valid trend direction is noted within the channel. If the value breaks out of the channel only then it becomes a "trend".

Hope this helps

Batchboy:Hi mladen,

Please elucidate more about this indicator. Does it start with nonLagMA? Is that how it originated? Then channeled with the slope value?

Then a 3rd component, a filter you say. Please elucidate more on how many components (3?) And what they do.

Thanks.mladen:It is a linear regression value "in disguise"

Whenever linear regression value slopes up value is added to "cumulative trend" value. When linear regression value slopes down that same value is subtracted from "cumulative trend" value. If the change of direction is within the channel width it is treated as an incomplete and undefined yet, only a previously valid trend direction is noted within the channel. If the value breaks out of the channel only then it becomes a "trend".

Hope this helpsThx :-)

You also mentioned you added a filter, can you tell me a little about?

BTW, 2nd parameter is "Channel Length", but it seems to adjust the width. Minor typo or I'm missing info?

...

You are right about the typo, but at the time of making it called it that way ...

As of filter : idea is simple - do not react immediately but wait some bars form trend change confirmation. If it continues int the same direction that it is a "new trend". Maybe the best is to look at the linear regression value colored on slope change and see what periods are tried to be avoided. Attaching one version that will show you a colored linear regression value on chart (the version that does not repaint) so you can compare it to the trend indicator and get a much clearer picture of what exactly is it all about

___________________________

PS: in the calculation I am using vinins shorter version of calculating linear regression value. He proved some 2 years ago that his way and the original way are mathematically identical and I like his way for the simplicity of it

Batchboy:Thx :-)

You also mentioned you added a filter, can you tell me a little about?

BTW, 2nd parameter is "Channel Length", but it seems to adjust the width. Minor typo or I'm missing info?Files:mladen:It is a linear regression value "in disguise"

Whenever linear regression value slopes up value is added to "cumulative trend" value. When linear regression value slopes down that same value is subtracted from "cumulative trend" value. If the change of direction is within the channel width it is treated as an incomplete and undefined yet, only a previously valid trend direction is noted within the channel. If the value breaks out of the channel only then it becomes a "trend".

Hope this helpsThx mladen,

You say wait number of bars, how have you set it to determine that, fixed? And maybe that should be external variable?

I'm finding 1min TF and lengthy channel (1000) with so-so narrow width like 35 to have excellent character, good trend follow with little indecision. So even if you did give variable on waiting not sure anything could be better.

Main question I have now is, how does it not take srious the medium whiplashes, the opposite accumulation doesn't reverse the channel yet? Yet when subtle moment arrives to look the other way, how is this possible

Old saying, if something seems too good to be true, it probably isn't true...... But I've not been dis-enamored yet,so its eating at me, what am I missing as far as the drawbacks?

...

Number of bars is already a parameter (the is the ChannelLength - as we said it should say width but let us leave it as is for now, the length is referring to the way how it is calculated internally)

Please, play a bit with the indicator parameters to explore how it works. Also compare it to the one posted a few posts ago (the linear regression value) in order to familiarize how it works. I am afraid that nobody can answer all the questions of some indicators behavior in some situations (otherwise its name would be the "holly grail") Experiment and use what indicators are best at : immediate visual result of some settings

regards

Mladen

Batchboy:Thx mladen,

You say wait number of bars, how have you set it to determine that, fixed? And maybe that should be external variable?

I'm finding 1min TF and lengthy channel (1000) with so-so narrow width like 35 to have excellent character, good trend follow with little indecision. So even if you did give variable on waiting not sure anything could be better.

Main question I have now is, how does it not take srious the medium whiplashes, the opposite accumulation doesn't reverse the channel yet? Yet when subtle moment arrives to look the other way, how is this possible

Old saying, if something seems too good to be true, it probably isn't true...... But I've not been dis-enamored yet,so its eating at me, what am I missing as far as the drawbacks?mladen:Number of bars is already a parameter (the is the ChannelLength - as we said it should say width but let us leave it as is for now, the length is referring to the way how it is calculated internally)

Please, play a bit with the indicator parameters to explore how it works. Also compare it to the one posted a few posts ago (the linear regression value) in order to familiarize how it works. I am afraid that nobody can answer all the questions of some indicators behavior in some situations (otherwise its name would be the "holly grail") Experiment and use what indicators are best at : immediate visual result of some settings

regards

MladenLOL, true!

I have been playing with. As far as experimental play I'm trying to get paid mql programmer to do a spec sheet I wrote for an EA that in optimization runs will do the ultimate "experimental play".

Thx!

mladen:doody

You are right. The indicator you posted does not repaint.

Here is the same thing but written differently (simpler code, no restriction in bars (it works for the whole chart now)) I think it will be clearer what does it do (from the code) now and it is more suitable for further working on now.

PS: kept all the logic same as in the original

regards

MladenHey mladen - this is pretty good if I set risk to 1. Could you or someone else filter most of these arrows away via changeable semaphor range/period ?

Files:Divergence Price and Volume Indicator

Hello,

i have found the following picture:

Divergenz-Sensor

Can someone make this Indikotor?

Thanks and Regards

derumuro

voltagetoe:Hey mladen - this is pretty good if I set risk to 1. Could you or someone else filter most of these arrows away via changeable semaphor range/period ?

Hi Voltagetoe,

Small change just made the x1 and x2 user controlled, the x1 is overbought for normalized Wpr and x2 is oversold for normalized Wpr. The indicator is already using range to determine arrow distance from price.

Files: