The Sultonov Regression Model (SRM) - claiming to be a mathematical model of the market. - page 8

 
faa1947:

Which is it going to be: supporting the man or dividing the kotir?
I think it is a thankless task to divide market quotidian into "noise" and "non-noise". Without "noise" there is no "non-noise", pardon the pun. We are not yet mature enough to separate these Siamese twins.
 
TheXpert:
Stop playing around. Sharing a kotir.

The prediction model is correct if the residuals are normally distributed.
 
TheXpert:
Stop cheating. Sharing the kotir.


Well, I didn't get it right away.

I have written many times. So it's not up to me to "yowl".

It's not my idea. We allocate what we can, step by step. In TA we take a mashka and work, but we forget about the difference between a mashka and a kotir. I, and a lot of other people with me, don't forget anything and keep modeling the rest, trying to refine the model. In a nutshell.

Have written and posted models many times.

 
TheXpert:
Stop playing around. Sharing a kotir.
Tell me the best way to divide. Do you want to feed the RMS inlet with a doughnut?
 
Avals:

The prediction model is correct if the residuals are normally distributed.


Well, more precisely, they're stationary.

What if they're not quite stationary? A little, a little, What if it's not quite, a little? Where's the boundary?

 
yosuf:
Suggest the best way to separate.

Plenty. I think your 18 is doing just that.
 
faa1947:


Well, more precisely, stationary.

What if they're not quite stationary? A little, a little, What if it's not quite, just a little? Where's the boundary?


Well, to be more precise, the residuals are still normal)))
 
faa1947:


Well, more precisely, stationary.

What if they're not quite stationary? A little, a little, What if it's not quite, a little? Where's the boundary?


No, exactly normal.
 
Avals:

the prediction model is correct if the residuals are normally distributed.
Where do we find a quotient with normally distributed residuals? We can only make it up, but we do not consider that case here.
 
yosuf:
Suggest the best way to separate. Do you supply the RMS input with a doughnut?
Faa, yosuf, essentially you need to split the original series into two components useful information and noise, and then feed the useful component to the RMS? Have I got that right?
Reason: