The potential yield of the instrument. - page 6

 
Avals писал(а) >>

as you correctly noted the average length tends to 2H - which is about as much as Hurst 0.5. Pastukhov and Shiryaev, for example, considered this measure (h-volatility) as a property of an instrument and the basis for deciding on its trading method.

But it is wrong to take the average knee value when analytically deducing the maximum earnings, because we are essentially allowed to tweak it. I.e. it is not obvious that the maximum amount will be described as a function of ZZ's average knee multiplied by the number of trades.

I agree that the correct solution is ZZ with spread or spread+1 and the difference will be in the form of trades with zero profit

But you are wrong! - Exactly ZZ with spread parameter, without +1.

Here are the numerical simulation results for the Wiener process (martingale) compared with the analytical solution found:

One can see a satisfactory agreement and an exact match of the maximum. So there is no contradiction, while your mql4com algorithm for constructions is most likely faulty. I won't search for it, using the presumption of innocence. It's you who should firstly prove that there is an error in the analytics and then prove that my algorithm of means is not correct.

 
Neutron, post your MathCad file (just save it in the format of the earliest possible version of MathCad. I made my file in MathCad6). I'll have a look at it.
 
mql4com писал(а) >>
Neutron, post your MathCad file. >> I'll have a look at it.

>> Catch!

Files:
zz_1.zip  328 kb
 
For ZZ with spread+1, minimum knee=spread+1 and minimum earnings=1. That said, all swings of length>=spread+1 with maximal rollback in spread - what else can be squeezed out of them? After all, ZZ nodes with spread+1 are a subset of ZZ nodes with spread parametre.
 
Neutron, your prices are not a multiple of a point.
 

How does this work? Here is the block in the body of the program that rounds the original BP to natural:

P.S. Saved in MathCad6.

Files:
zz_2.zip  317 kb
 

Adjusted the Matkad file so that it would swallow the cotier rounded to whole numbers.

Files:
zz1_1.zip  141 kb
 

Neutron, you have an error in the ZigZag:

ׂ

With a spread of 19:


 
I'll have a look.
 
mql4com писал(а) >>

Neutron, you have a bug in ZigZag:

That's it, mql4com, you got me - there really is a bug in the code! Gee, I was looking at it...

I'm off to clean up.

>> Thanks.