From theory to practice - page 612

 
Evgeniy Chumakov:
Yes the question is how to pull those ears to make it work, but it's just a pretty picture, so far I've put it in a drawer.

Yeah... I'm sitting here looking for information on such distributions, where they occur and what I can get out of them...

In the meantime, still running my TS in the =24 hour window. The trump card of a near Poisson flow of quotes smoothing out intraday volatility beats all other arguments.

 

Here is another series of the sum of the increments in the 1440 minute window over the 40000 bar M1.

The price is pre-filtered here, I don't know what happened.

I still can't make a channel, something is wrong.

 
Evgeniy Chumakov:

Here is another series of the sum of the increments in the 1440 minute window over the 40000 bar M1.

The price is pre-filtered here, I don't know what it is.

Still can't build a channel, something is wrong.

No, it's a total junk...

I stand by my opinion - either I simply work with increments in the sliding window = 24 hours and look at correlation coefficients or Hearst's coefficients when the sum of increments exceeds the channel limits (i.e. go to the classical Herald), or show my genius and learn to work with "forex ears".

 
Alexander_K2:

Nah, that's a load of crap...

I stand by my opinion - either to work stupidly with increments in sliding window = 24 hours and look at correlation coefficients or Hearst coefficients when sum of increments exceeds borders of the channel (i.e. to go the severe classical way to the Grail), or to show genius and learn to work with "forex ears".


Thanks for the histogram, one can see that the idea needs to be scrapped. Another thought did come up though......


If you remember the screenshots of trades (I posted the last ones), where the entry is right from the top of the candle.... I was testing these ears, but as always there are wrong entries, which you need to sift out.
 

Probably it is necessary to make incremental increase of forecast error, i.e. by some method forecast forward is made, then we compare result and forecast and have let's say forecast error, distribute errors, etc. Exit outside of the channel limits signal that probability of error of the next forecast is small > entry in a deal.


This is just me writing fantasy stories))))

 
Evgeniy Chumakov:


Thanks for the bar graph, you can see that you need to scrap the idea. Another thought did come up though......


If you remember the screenshots of trades (I posted the last ones), where the entrance is right from the top of the candle.... I was testing these ears, but as always there are wrong entries that need to be sifted out.

Yes, there was a class there.

How do you sift it? I don't think there's any other way but to look at correlation or Hearst. Although, Hearst does not work for me at all, and correlation is 50/50, but there is nothing else to look at from classical parameters "trend/flop"...

Without additional parameter my profit is strictly =+0%. Without this parameter there is nothing to do in the market - this has already been proven on my deposit.

 
Alexander_K2:

Nah, that's a load of crap...

I stand by my opinion - either I work in a dumb way with increments in a sliding window = 24 hours and when the sum of increments exceeds the channel limits, I look at correlation coefficients or Hirst (i.e. take a classical path to the Grail), or ask for genius and learn to work with "ears of Forex".

How do you know the classical way to the grail? ))

 
Alexander_K2:

Nah, that's a load of crap...

I stand by my opinion - either I work in a dumb way with increments in a sliding window = 24 hours and when the sum of increments exceeds the channel limits I look at correlation coefficients or Hirst (i.e. take a classical path to the Grail), or ask for genius and learn to work with "ears of Forex".

to the eye - complete chaos, reminiscent of increments in an exponential reading of quotes.

 
danminin:

and how do you know the classical way to the grail?)) eh? ))

Bass and Asaulenko went down it. Hard workers... From the farm, so to speak... Well, me too - I don't have enough insight or something. Dumbed down, in short.

 

And what does the histogram show here.... This is what the sum of the increments looks like.


Files:
Reason: