Testing real-time forecasting systems - page 25

 
forte928 писал(а) >> Hondrick-Prescott filter

Overdrawn....

 
LeoV писал(а) >>

Overdrawn....

when there are sharp fluctuations and the lambda coefficient is high...

If lambda coefficient =3 then in the absence of sharp fluctuations the plot is reliable and this curve can be trusted...

 

very nice forecast, look at it as a picture or as a trading recommendation.... it's up to you


 

Before Njel resumes publishing FX13 forecasts, I will post pictures of his developments:

and a couple more here.

 

Yeehaw, that went "usd-jpy" well.


 

It's getting late :o) Let's shake off the old days. The model is raw, so just to be on the safe side:

Not for trading

EURUSD

H1

Forecast for the week ahead (120 counts) starting this Monday (zero count is the first trading hour of Monday):



Pure statistics, let's see what happens. :о) True, there is still an alternative option, and fuck it. You can see in the picture (hopefully it's already clear who's where):

  • the most probable boundaries of the process development
  • specification of the process implementation, to identify possible reversal zones

I'm not giving more details with calculation of trading levels, it's still not clear whether it will work or not.


PS1: does anyone have something similar?


PS2: I should move it to MT for the sake of convenience, but it is too much for me :o(

 
grasn >> :

PS1: does anyone have something similar?

>> I do.


 
Ko1dun >> :

>> Got it.


>> conceptually similar, let's see :o)

 

"The hardest part of our job is waiting" (C)


 
Ko1dun >> :

Bingo.


What's that?

By the way, our forecasts are the same .....