Triangular moving average (TMA) ... - page 6

 

I've got a feeling, a feeling deep inside, oh yeah! - Beatles

Yes, I'm old! ;^) Same age as the Beatles, pretty much.

Would you mind answering a question for me? I wonder if all the descriptions of TMA are equivalent?

For example, you referred to a description of TMA on About.com in post #2. Baluda (stevehopwood forum) on the other hand uses one that seems different (but which I understand better ;^), but it could be equivalent.

stevehopwoodforex.com • View topic - Baluda Indicators

Baluda:

"...So the formula in simple form:(21 * Close[0] + 20 * Close[1] + 20 * Close[-1] + 19 * Close[2] + 19 * Close[-2] + ...) / (21 + 20 +20 +19 +19 + ...)

This is the TMA, a double smoothed average of the closing prices of bars to the left(past)and right(future)....

The TMA and Slope for the last 20 bars changes with time."

This doesn't seem to agree with this description for About.com: "...the additional smoothing actually moves the domainant input value to the middle of the input series (which would decrease responsiveness)."Except for the most current 20 price bars, Baluda's method appears to be heavily weighted toward the bar for which the TMa is being calculated, not the middle of the series, since it has the highest multiplier and the previous 2 bars would be next highest, etc.

Of course, the most current 20 bars will have a TMA value calculated in a truncated way, ranging from missing 1 'future' bar (the 19th bar) to missing all of them (the current bar). Not sure that would shift the dominant value to the middle of the input series, though....

Is Baluda's version valid?

 

...

Baluda is actually describing a "centered" triangular moving average not a "regular" TMA

Here is one more link with even simpler explanation (formula) of "regular" triangular moving average : Investor/RT Tour - Moving Averages

Quinn Ronin:
Yes, I'm old! ;^) Same age as the Beatles, pretty much.

Would you mind answering a question for me? I wonder if all the descriptions of TMA are equivalent?

For example, you referred to a description of TMA on About.com in post #2. Baluda (stevehopwood forum) on the other hand uses one that seems different (but which I understand better ;^), but it could be equivalent.

stevehopwoodforex.com • View topic - Baluda Indicators

Baluda:

"...So the formula in simple form:(21 * Close[0] + 20 * Close[1] + 20 * Close[-1] + 19 * Close[2] + 19 * Close[-2] + ...) / (21 + 20 +20 +19 +19 + ...)

This is the TMA, a double smoothed average of the closing prices of bars to the left(past)and right(future)....

The TMA and Slope for the last 20 bars changes with time."

This doesn't seem to agree with this description for About.com: "...the additional smoothing actually moves the domainant input value to the middle of the input series (which would decrease responsiveness)."Except for the most current 20 price bars, Baluda's method appears to be heavily weighted toward the bar for which the TMa is being calculated, not the middle of the series, since it has the highest multiplier and the previous 2 bars would be next highest, etc.

Of course, the most current 20 bars will have a TMA value calculated in a truncated way, ranging from missing 1 'future' bar (the 19th bar) to missing all of them (the current bar). Not sure that would shift the dominant value to the middle of the input series, though....

Is Baluda's version valid?
 

A helpful modification?

If Baluda's way of calculating TMA is correct, has anyone considered calculating the last 20 bars (for a 20 period TMA) the following way?

Instead of truncating one side of the calculation for the most current 20 bars (due to lack of 'future' bars), the period of the TMA calculation would be shortened until there were enough 'future' bars to make it balanced.

Calculate the 20th bar back in the regular fashion (there are enough bars for the 20 period TMA)

Calculate the 19th bar back as a 19 period TMA

Calculate the 18th bar back as a 18 period TMA

Calculate the 17th bar back as a 17 period TMA

.

.

.

Calculate the 3rd bar back as a 3 period TMA

Calculate the 2nd bar back as a 2 period TMA

Assign the current bar it's closing value.It seems like it might create a smoother transition from the non-truncated TMA to current bar's price. It looks like it would be a better balanced transition, but what do I know? ;^)

I'd love to see how a TMA with this modification compares to the standard one. If you know of one that's calculated this way, would you please let me know?

The following version (of your original, I imagine ;^) was supposed to demand less resources to use.

Would it be possible for you to modify it in this fashion? I'd be very grateful if you can, and very grateful if you can't! ;^D

 

...

Quinn Ronin

Please read the previous answer. I will just repeat that that is not TMA describtion, but centered TMA description

regards

Quinn Ronin:
If Baluda's way of calculating TMA is correct, has anyone considered calculating the last 20 bars (for a 20 period TMA) the following way?

Instead of truncating one side of the calculation for the most current 20 bars (due to lack of 'future' bars), the period of the TMA calculation would be shortened until there were enough 'future' bars to make it balanced.

Calculate the 20th bar back in the regular fashion (there are enough bars for the 20 period TMA)

Calculate the 19th bar back as a 19 period TMA

Calculate the 18th bar back as a 18 period TMA

Calculate the 17th bar back as a 17 period TMA

.

.

.

Calculate the 3rd bar back as a 3 period TMA

Calculate the 2nd bar back as a 2 period TMA

Assign the current bar it's closing value.It seems like it might create a smoother transition from the non-truncated TMA to current bar's price. It looks like it would be a better balanced transition, but what do I know? ;^)

I'd love to see how a TMA with this modification compares to the standard one. If you know of one that's calculated this way, would you please let me know?

The following version (of your original, I imagine ;^) was supposed to demand less resources to use.

Would it be possible for you to modify it in this fashion? I'd be very grateful if you can, and very grateful if you can't! ;^D

 

Sorry, I posted it before I read your answer.

mladen:
Quinn Ronin

Please read the previous answer. I will just repeat that that is not TMA describtion, but centered TMA description

regards

Sorry. I've been up all night, and I'm a tad punchy. I wrote all three post together, but they were too long to post, so I broke them up. I didn't expect such a prompt reply. I'll know next time.

The description on the site you gave me was very clear, and the TMA is obviously very center weighted.

I also noticed looking at the various moving averages that the least square, the endpoint and the modified appear to have the least lag, with the least square being the smoothest of the three.

Have you done anything with the least square MA?

 

...

As usual that indicator has multiple names on metatrader Anyway, here is one version of it : lsma nrp

Probably the proper name should be linear regression value (and I prefer that name) but LSMA will do too

Quinn Ronin:
Sorry. I've been up all night, and I'm a tad punchy. I wrote all three post together, but they were too long to post, so I broke them up. I didn't expect such a prompt reply. I'll know next time.

The description on the site you gave me was very clear, and the TMA is obviously very center weighted.

I also noticed looking at the various moving averages that the least square, the endpoint and the modified appear to have the least lag, with the least square being the smoothest of the three.

Have you done anything with the least square MA?
 

Thanks!

Much obliged, mladen. I'll give it a whirl, though TMA has been my favorite for several years. So smooth and responsive at lower periods.

Parabolic SAR is an indicator that isn't calculated like the others, is it? I've been working on using multiple SAR's sort of like MA's. I finally found a way to set the settings that allows me to adjust the distance from price easily. Simple when you know the trick, but murder to set by trial and error!

 

Wow!

mladen:
As usual that indicator has multiple names on metatrader Anyway, here is one version of it : lsma nrp Probably the proper name should be linear regression value (and I prefer that name) but LSMA will do too

That LSMA is fast. Much faster than the TMAs I've been using. I like the color change as well. Tks!

 

...

A good thread with some more versions and some more information on linear regression value (or lsma) : https://www.mql5.com/en/forum/174318

Wish a nice weekend to all

Quinn Ronin:
That LSMA is fast. Much faster than the TMAs I've been using. I like the color change as well. Tks!
 

"Lwma" slope with alerts ...

Its is a so called "lama" slope with alerts (since it is not a lwma but something similar to some extent to it). Posted it here since it was discussed at this thread what TMA is and what it is not and the original request for it came from something called "TMA slope"