Comparison of two quotation charts with non-linear distortions on the X-axis - page 4

New comment
 
hrenfx: It draws correctly a classic ZZ - someone else's invented indicator that fails to even show the best inputs/outputs on history.
Yes, it is not difficult to change the code to apply a different ZZ, on the previous page I uploaded the ZZ angles to the file, so far I see a difference in statistics for ZZ tops up and down, I have not yet analyzed for the minute TF, but I think there will also be different statistics for up and down corners
 
hrenfx:
  1. The original TsVRs (Bid and Ask) are logarithmised.
  2. ZZ is constructed with the condition that the knee size >= N and the number of knees is maximal. Top tops lie on Bid, bottom to Ask.

Did not check for symmetry criterion.

You have explained several times the need to logarithm the original BPs, if I don't confuse anything, this is necessary to bring the cvr to the same view, right?

if not, can you explain?

And I don't understand at all, how is logarithmation of tsvr done???

Thank you.

 

Consideration of absolute price change is illogical for several reasons:

  1. The change in the price of a pip for a single FI over time. For example, gold 5 years ago and now.
  2. The impossibility to compare two Instruments with each other. For example, a 1 pip change of EURUSD and GBPUSD are completely different events.

Relative price change does not have these disadvantages. Logarithm of MPT allows to pass from a relative estimation to the absolute one (simpler to implement): log(a / b) = log(a) - log(b).

Logarithmic IRR - each term of the series is equal to the logarithm of (any) price: Prices[i] = Log(Prices[i]).

 
hrenfx: The logarithm of the qVR - each term of the series is equal to the logarithm (any) of the price: Prices[i] = Log(Prices[i]).

what can it do?

here is the euras prologarithmic H4 http://imglink.ru/pictures/07-03-12/5735ba01c1d14f9baa96ad92c1afc800.jpg

and a normal chart: http://imglink.ru/pictures/07-03-12/1ad651a678ac73189f5ada780de6508c.jpg

to be honest, I don't see any difference

 

There are tasks where not having logarithms won't make a difference. But there are tasks where it is practically impossible to do without it:

  1. Searching for relationships between different FIs.
  2. Searching for similar sections of history in the same FI, with a deep history.

From the academic point of view, logarithm must always be done. From a practical point of view, it is by no means always necessary.

 
hrenfx:

Consideration of absolute price change is illogical for several reasons:

  1. The change in the price of a pip for a single FI over time. For example, gold 5 years ago and now.
  2. Impossibility to compare two Instruments with each other. For example, a 1 pip change of EURUSD and GBPUSD are completely different events.

A relative price change does not have these disadvantages. Logarithm of MPT allows to pass from a relative estimation to the absolute one (simpler to implement): log(a / b) = log(a) - log(b).

Logarithmization of TDT - each term of the series is equal to the logarithm of (any) price: Prices[i] = Log(Prices[i]).


excuse me, the first and fourth lines don't contradict???

from the first one we understand that the absolute change is illogical and relative change should be used.

from the fourth, logarithm allows you to go from relative to absolute.

confused.

 
After logarithm, the absolute estimate is equal to the relative estimate before logarithm.
 
Thank you.
 

Scriptong did something like that, I'd have to check his forum


Files:
nextbartype.mq4  23 kb
 
Integer:


Take the zigzag thicker. What is DTW? Is it just a scaling along the time axis? But it's uniform. If it's a satisfactory solution, then there shouldn't be any question, it's the first elementary solution.

Simple and beautiful method, I'm extremely surprised I haven't encountered it before.

As I understand it, you take two signals of the same length, build a matrix of pairwise distances between samples and use it to find the "lowest" path from the lower left corner to the upper right corner. It is taken as the required non-linear time transformation:

The picture is from here, there is also a detailed explanation of all the nuances. The code in mql will probably not be long))

123456789101112
New comment