stevehopwoodforex.com
https://www.stevehopwoodforex.com/phpBB3/
Print view

TMA Discussion Thread
https://www.stevehopwoodforex.com/phpBB3/viewtopic.php?t=587		 Page 6 of 17
Author:  garyfritz [ Thu Jun 28, 2012 1:51 pm ]
Post subject:  Re: Don't trust visual backtests with TMA slope
Jemook wrote:Ok all tma discussion is moved here now. Id like to rename the thread to TMA Discussion Thread but cant work it out. Steve any ideas?
I renamed it. Just had to edit the first post in the thread. (Which was no longer the first post in the original "don't trust visual backtests" thread?)
Author:  Jemook [ Thu Jun 28, 2012 2:25 pm ]
Post subject:  Re: TMA Discussion Thread
Good stuff, continue. I'd like to hear bobs thoughts on the new indicators we are working with which use 'tma true' settings.
Author:  garyfritz [ Thu Jun 28, 2012 2:49 pm ]
Post subject:  Re: TMA Discussion Thread
If I understand Bob's objections, he's concerned that the "true" variants don't provide the same values in realtime as the original future-peeking TMA. That would indeed be a problem.

But there's no reason it has to be different. The future-peeking TMA, by definition, can't peek into the future on bar 0. So if you write an indicator that does exactly what TMA does on bar 0, and then does the same on ALL bars, then you should have exactly the same value on bar 0 and the correct non-modified values on historic bars. I think that would meet Bob's needs and it would avoid the confusion of the repainting indicator.
Author:  NeoTrader [ Thu Jun 28, 2012 3:38 pm ]
Post subject:  Re: TMA Discussion Thread
hi gary, jemook
garyfritz wrote:If I understand Bob's objections, he's concerned that the "true" variants don't provide the same values in realtime as the original future-peeking TMA. That would indeed be a problem.
Bobs objections are not about the values of the current bar...since this values are the same on both versions...
Bobs objections are about the range in which the normal TMA operates and the range in which the TMAtrue is operating. And that his Levels of 0.4 and 0.8 could be off in the true Mode since the range scale is different (the values at the right side means the scale of the range that is outside the right side of the indicator in the vertical area where also the price values from the chart reside).

I also had the idea that he was talking about the actual values of the current bar which i didn't understand since they where always the same in both modes.

If you look back a few posts I also understand the first time what he really meant and made some points regarding his objections about the range...and bob seems to be open to this points and is going to look into it...

Please check the last posts from me and bob about this. Than you will also understand what his real objections are...

Any insights from your side would be highly appreciated by me since I think we both are the major opposition against this repainting nonsense.


happy trading,

NeoTrader

-
Author:  garyfritz [ Thu Jun 28, 2012 3:57 pm ]
Post subject:  Re: TMA Discussion Thread
If the bar-0 values are the same -- which you've shown they are -- then how could the range be different? If the realtime bar-0 result is the same for "classic TMA" and "true non-peeking mode," then the 0.4 / 0.8 / etc values HAVE to be the same on bar 0. Bob's bar-0 trading logic should be unaffected.

If the bar-0 values are the same, the only way I can imagine the right-side range scale being different is if the smoothed "classic TMA" has a larger or smaller range than the non-smoothed "true" mode. The right-side range scale is controlled by all the histo bars on the screen, not just the bar-0 value.

Could you post a picture of the two indis with different scales so we can be sure we're talking about the same thing?
Author:  NeoTrader [ Thu Jun 28, 2012 5:47 pm ]
Post subject:  Re: TMA Discussion Thread
hi gary,
garyfritz wrote:If the bar-0 values are the same -- which you've shown they are -- then how could the range be different? If the realtime bar-0 result is the same for "classic TMA" and "true non-peeking mode," then the 0.4 / 0.8 / etc values HAVE to be the same on bar 0. Bob's bar-0 trading logic should be unaffected.
exactly my thoughts...and TMAtrue also works with his levels as far as I have experienced till now.
garyfritz wrote:If the bar-0 values are the same, the only way I can imagine the right-side range scale being different is if the smoothed "classic TMA" has a larger or smaller range than the non-smoothed "true" mode.
I had also the same assumption and described it to bob in one of the former posts...and also that the fictional values of the standard TMA may be responsible for a larger range than in true mode...since in this mode only real time data is shown...and that the range of the true mode is more accurate than in standard mode.
garyfritz wrote:The right-side range scale is controlled by all the histo bars on the screen, not just the bar-0 value.
yep this is correct ... all bars you can see in the indicator (chart) at that current segment influence the scale...
garyfritz wrote:Could you post a picture of the two indis with different scales so we can be sure we're talking about the same thing?
Your wish is my order.. ;)
I marked the scale values...but I think that we are finally talking now about the same thing here...;)
tma slope range difference.png
As you can see the scale of the standard TMA is about 60% to 100% larger than the true mode...

And one thing more... The true mode still repaints slightly. But on a much much smaller scale than the original one...barely seeable. Just put them in the strategy tester (visual mode) and follow them...you will see that there are still some repaints but really not in a relevant range...and far from changing the values as much as the original TMA.

And also as a sidenote this TMAtrue mode is the original refined mode that was developed by zznbrm and is used in all current indicators from baluda (ignoreFuture Parameter) and me (tmaTrueMode parameter).


happy trading,

NeoTrader

-
Author:  garyfritz [ Thu Jun 28, 2012 7:21 pm ]
Post subject:  Re: TMA Discussion Thread
OK, the bar-0 values are identical. The historic-bar values on the "true" indi should be the same (or very very close) as they were when they were at bar 0. Clearly the fictional repainted values of the standard TMA greatly exceed the original bar-0 values. I haven't looked at the code deeply enough to understand why, but I don't think "why" is important. The larger scale on the standard TMA is as fictional as the smoothed values that caused it. The "true" indi is displaying reality, present and past.

Have you looked at the "true" indi code to see why it's repainting at all? If it does a pure LWMA (which is basically the left half of the triangular MA on bar 0) then it shouldn't change a bit.
Author:  Baluda [ Thu Jun 28, 2012 7:38 pm ]
Post subject:  Re: TMA Discussion Thread
Gary,

Please check this post. It explains all calculations.

Paul
Author:  garyfritz [ Thu Jun 28, 2012 7:46 pm ]
Post subject:  Re: TMA Discussion Thread
Yes, I saw that. It doesn't tell what the "true" mode does and why it would modify its values over time.

You might be amused to know that I had no idea what the "TMA" in this new 10.2 indicator stood for. :oops: I finally read the code and thought it looked like a LWMA but with this crazy bit that looked forwards AND backwards. It wasn't until you mentioned in that post it was a Triangular MA that I had my "aha" moment. :D
Author:  NeoTrader [ Thu Jun 28, 2012 8:03 pm ]
Post subject:  Re: TMA Discussion Thread
hi gary,

you can also take a look at this post where baluda shares the complete code of both calculations.

happy trading,

NeoTrader

-
All times are UTC Page 6 of 17

Powered by phpBB® Forum Software © phpBB Limited