How to form the input values for the NS correctly. - page 16
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
You can try the entry points of your strategy
if these points are perfect - there's nothing else needed - no nets!
>> right?
I'm talking about points - where it makes sense to teach the network to DIVERSE
I was going home and thought about how I would propose to input zz extremums... I did not have time... I propose to input 4 confirmed values of WP and predict the next value, normalize extrema by EMA, for example 120, for 1-hour charts
I think it would be better to feed the sample with a somewhat formed reversal, rather than immediately against the movement at the peak.
I think it would be better to submit the pattern with a somewhat formed reversal, rather than immediately against the movement at the peak.
The beginning of the 2nd wave is ideal.
as a rule it is the first covergence on the calculated period
It seems to me that it would be better to submit a sample already with a somewhat formed reversal, rather than immediately against the movement at the peak.
Decide on the criterion for defining pivot points and you'll get points. What is flat for one is a trend for another.
What is there to determine
these points are different for each period (tf)...
on W1 - trend on M15 may be trend - on D1 - flat
---
first you have to determine the period
---
there is no point ---
one trader jumps on M1, the other on W1 - MN1
one catches 10-20 pips - the other takes weekly candles
they have the same points - no of course not
---
What do you mean a shape? I suggest a strategy based on a zz, assuming a certain mutual location of zz extrema leads to something and then using book methods - since the gurus have not yet responded - to pick the best transformation
no, not the figure, I mean.... this one... learning vector.
We can formalize it this way, if the next extremum is between two previous extremums - flat (0), above the peak - trend is up (+1), below the trough - trend is down (-1), we forecast -1 0 +1, we should start somewhere... If we change the way of normalization of input data we will choose the least error in the forecast
A little mistake, we may have either breakout of the last extremum or formation of an extremum in the range of the previous two extremums and we will forecast 2 market conditions....
The very strategy of placing a pending order-stop at the last confirmed extremum produces a normal profit at the euro watch from the middle of 2007 omitting December 2007 and January 2008. Using the above mentioned NS we can filter out the signal to place the order
... by changing the way the input data is rationed
Exactly, if you can expand on this point, because it is the most unclear.
What rationing means in general. And how to normalize so that in the future the range was in the 0-1 range and we don't care about the values of initial input non-normalized data.
On which sample should we normalize (all samples or just the current one)?
Which view (lined or function of the s-species).
If linear, the range in the future 0-1 may not be saved (a value greater than 1 will be found) and the network will not be trained on it.
If s-species, there is a saturation of the large ones and they will no longer be distinguishable for the network.
Is there a golden mean?