Digital ACSTrend - page 20

 
Big Joe:
Oh Well Can you please, post your work please ? Very interesting !

+1

I would like to look at.

 

More function in Auto-ssa

I found Dr. Theodore Alexandrov (autossa@gmail.com), auto-ssa matlab version modified: 6 Oct 2006.

There are more functions in auto-ssa, eg, fourier identify, auto periodicity, LF identify, periodogram, reconstruction, auto trend, period_est.

But, you have to study matlab programming language to get all stuff out.

 
newdigital:
Hi John Last,

I already posted about this commercial software namely Caterpillar in 2005. See this thread:

https://www.mql5.com/en/forum/173114

With indicator for MT4.

It was in 2005. Now is 2011.

Can you imagine how old I am?

It is from the same series and seems - made by same people

Welcome to AutoSSA!

2011 - 2005 = 6 ... at least

 
Jack1:
I found Dr. Theodore Alexandrov (autossa@gmail.com), auto-ssa matlab version modified: 6 Oct 2006.

There are more functions in auto-ssa, eg, fourier identify, auto periodicity, LF identify, periodogram, reconstruction, auto trend, period_est.

But, you have to study matlab programming language to get all stuff out.

Thanks Jack, that's great !

Can you also share your modified trend magic indicator, this will be super cool !

Thanks again

 

Thanks for the link

Thank you New Digital for the link.

Indeed 6 years is a lot of time.

There may exist some periodicities in the systems LOL.

 

Just some clarification :

"Regular" SSA will recalculate last Lag bars in every situation when a current value is changed (it is its "nature" and it is the way it is calculated). So if the Lag is > 1 (any value, you can put 1000 for Lag too and it will change all the 1000 bars, the fact that those changes are small does not change the fact that it will change them) So far I have not found any situation when a current bar value change would cause a situation when SSA does not change previous Lag values. The purpose of end-pointing was to avoid that

The current valueof the 2 (end-pointed and the "regular") is the same and it must be the same (that is the second significant property of the end-pointed version - it is SSA after all, and if we had different values for same inputs than what would one of it be?)

__________________________

And the last : there is no difference in values when using libSSA.dll and SSA.ex4. The difference is entirely in different code and in speed of execution. I already explained in a couple of places that I made libSSA.dll because a friend of mine asked me if something can be done regarding the speed of SSA calculation, and that is the result. Elite section members have the access to previous dll version C source and they can tell which are the differences (or simpler, since there are fewer of them, which are the similarities in code) As far as the current source code is concerned I did not release it, but I would remind that the values of both must be the same regardless of the way they are calculated

Hope this clarifies some things

Jack1:
Hi, Joe,

I did test both: SSA.ex4 and libSSA.dll. I did swap off them. Both produce the same result. No difference at all.

I add a procedure(some code) in my ssa, which reduce re-painted bar, then, it is no repainted indicator. I attached a chart here. You can validate if there are any big difference between them. They produce the same result for last bar: 0.2263.

My coding is very bad, too many error, un-reliable. So, I generally don't release my code to protect public. But, Mladen's code is quite good and safe. So, everything is ok for study, research, trial, etc. to work out this ssa theme.

Best Regards.
 

New concept of trend: SSA trend.

Best Thank Mladen, Your points are perfect right, wonderful explanation.

Due to SSA do a huge math data processing. A little change of input will produce a little difference in result. But, it won't affect trend extraction by SSA. So, no worry, SSA is a good, powerful tool to use.

Big problem is "don't know which is the best way to use SSA". I put out a chart here, which may help you to grasp this new concept and parameter setting.

Cheer!

 

Another market condition template

Here I add another market condition template.

We have:

1. Bollinger bands versus Fractional bands.

When the fractional bands exceed the Bollinger we can approximate that the market is no longer approximated by the Gaussian distribution.

The Fractional bands give a nice visual representation when a fractal break - out occurs.

(it may be used as a system by itself)

2. We have Hurst Difference indicator

Here we have a measure of the change of the Hurst exponent. When it is positive we have an indication that not only that we have a bigger probability (provided that we have a low fractal dimension) that the next bar will be in the same direction than the previous but this probability is increasing. (The Hurst exponent (Fractal dimension) does not gives big probabilities, when we have a 53 % probability it is very good).

We have the JCFBaux: This is a classic replacement of the ADX measure of the Trend. I use two plotted in the same window:

-period 15: measure of the short term trend

-period 50: measure of a medium term trend

3. Fractal Wave -PM. Camaron will explain better.

4. iVAR - fractal dimension

5. FGDI also fractal dimension but uses a different method.

The basic idea is that we need a template giving us a picture of the market state. We receive this picture measuring some market characteristics that cannot be estimated by the naked eye, e.g. the fractal dimension characteristics.

The idea is not only to measure the market state but also to spot as soon as possible the transition form one state to the other. Of course this is a modular template everyone can remove or add what he estimates worth. But I feel a necessity to separate directional indicators from market state indicators.

The idea is that the market state template would be universal whatever system we may apply. I was wondering is there an indicator that is really sound, and indicator that I can count whatever the market is: for example 100 period of SMA, or 20 period SMA, something really sound. Unfortunately there is not a such thing, so I am looking for something sound not in the directional indicators family but in the market states indicators.

Any comments or critics are welcome.

Files:
mc_t.gif  30 kb
 
Files:
orders.png  23 kb
mc_2.gif  27 kb
 
Jack1:
Hi, Joe,

I did test both: SSA.ex4 and libSSA.dll. I did swap off them. Both produce the same result. No difference at all.

I add a procedure(some code) in my ssa, which reduce re-painted bar, then, it is no repainted indicator. I attached a chart here. You can validate if there are any big difference between them. They produce the same result for last bar: 0.2263.

My coding is very bad, too many error, un-reliable. So, I generally don't release my code to protect public. But, Mladen's code is quite good and safe. So, everything is ok for study, research, trial, etc. to work out this ssa theme.

Best Regards.

Can you please, post your work please

thanks in advance