Stereo Neuro Net - page 5
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
It reminds me of a saying. Well, piss off, start from the beginning.
I'm sick of repeating myself. I wrote it all down on the previous page. You can look for a canal, flatten ..... flag on the play.
What do you mean first? I write the truth.
Here is the figure. Depending on the amplitude of these fluctuations, three market states can be distinguished:
Example:
1) amplitude of 18 pips - flat
2) amplitude of 82 pips - channel
3) amplitude of 567 pips - trends (succession of ascending and descending trends on the chart)
Generally speaking, a normal market condition is a flat. A channel happens when one or more big market participants open/close positions (they do it in batches, which leads to specific price fluctuations). A trend is caused by a very large exchange over a long period of time, usually either by long-term investments or the result of foreign economic activity (export/import).
what you're talking about, I understand perfectly. I wish you could understand me.
You have philosophical notions. Give me the formula. No formula, nothing to talk about. You can't put philosophy into a program.
If you think 83 pips is a trend and 82 is a flat. That's fine. That's the best definition of a trend I've ever seen :-). Explain these concepts to someone who works with daily candlesticks, and then prove the same to a scalper. Everyone is looking at this curve, through their TS. This curve doesn't care about your TS. It does not care what numbers you put there and what kind of logic.
The movement always has a direction and it is determined by the first and second derivatives at least. Trend = upwards or downwards directional movement.
The curve on the screen is MOVING. And it is only in our imagination that there is a flat, a channel and even that in hindsight.
Z.U. From this picture on the 100 count. How to determine the flat, the channel, in what state we are now ? How do we calculate it ? Are we in a channel or not ? Are we in a flat or not?
After any of your statements I am ready to draw a picture that will disprove your assumptions, according to the gradation you gave.
I fully share Prival's point of view - for some there is a "trend" and for others a prolonged "flat"! If we are even more categorical, the market does not have these states at all, everything depends on the TF in which a particular TS works. From this point of view, the picture can be treated as follows: for the TS working on TF=1, this is clearly a trend market. For the same TS, but working on TF=20 - it is a flat. I repeat (especially for budimir): Trend and flat are not categories that determine the market situation, this is a category characterizing a particular TS. I think this question cannot be explained easier and more clearly.
As for the method of viewing stereo pairs, you can get the methodology from the Internet by entering the keywords "stereo pair" or "stereo photo" in any search engine. I shall remind only that in the mid-90s there were on sale colourful albums with "latent" pictures, appearing only when properly managed to relax muscles of an eyeball and to combine two similar images in one. The same technique is used here as well, to construct a three-dimensional image.
budimir, there are no specifics about this NS - it is in a state of study and experimentation. I warned in the first post that you can't take the presented material seriously - it's not of practical interest. But what we are discussing here apart from that is useful and interesting.
Prival, in the demolished thread I responded to your comment regarding the state of TC "off market". The point comes down to the fact that this state is characteristic of a TS when there is a lack of useful information or it is contradictory. From textbooks it is known that in a situation where any movement is taxed (or punished), the optimal behavior for the Expert Advisor is to keep the current state unchanged (the principle of minimal resistance). Indeed, if the TS closes a position and then opens a new one in the same direction due to inconsistent data, we will lose on the DC commission compared to the case if we do nothing. In the case of opening in the opposite direction we have nothing to lose and nothing to gain. I.e., an Expert Advisor with binary output - Buy or Sell is optimal. The third one is redundant!
If you think 83 pips is a trend and 82 is a flat. That's fine. Best definition of a trend I've seen :-). Explain these concepts to someone who works with daily candlesticks, and then prove the same to a scalper.
Read carefully: i wrote "example"
Of course it's through the prism of TC that everyone is defined. But there have to be "eternal" principles, haven't there?
Give me the formula. No formula, nothing to talk about. ...
Dear Prival ,
and yet there is a specific mathematical formula for determining FLET:
We take the input attributes in the NS, with a straight face multiply them by the weights,
all that stuff runs through the synapses, then, with an even more honest face.
multiply by non-linearity (choice - like in a restaurant, according to taste)... and so on through
several layers, and the output is a beautiful neuron called FLET( FLAT, FLUT,
FLIT, FLOT-name you can choose to your liking) and if it activates, it means,
w-r-e-m-e-n-n-o-t.
to Neutron
В авишке, если правильно скосить глаза и впасть в состояние Нирваны, можно заметить, как 3-х слойная двух-входовая нелинейная сетка лопатит входные данные (ценовой ряд) пытаясь найти в них скрытые закономерности. И, таки, находит.
Seryoga hello. Help me out - I can't get my eyes right. Where's the pattern? Is that what "sticks to the corners" at the end of the cartoon? What's sticking to the corners? Unless, of course, it is a trade secret.
To understand the stalemate of it all - try to think of a winning strategy for playing chess ... with yourself. For money. :))
You're trying to catch something that doesn't exist, or rather what you have yourself.... :))
Amazingly put!
The mathematical reasoning here
to Neutron.
Seryoga hello. Help me out - I can't get my eyesight right. Where is the pattern? Is it that "sticking to the corners" at the end of the cartoon? What's sticking to the corners? Unless, of course, it's a trade secret.
What the hell kind of secret? - After Castaneda's revelations, the mystery is translated into practical terms. What matters here is the right source material and the skill of its preparation. Everything will work out :-).
Give me the formula. No formula, nothing to talk about. ...
budimir wrote(a) >>
Dear Prival ,
and yet there is a specific mathematical formula for the definition of FLET:
We take the input attributes in the NS, with an honest face we multiply them by the weights,
the whole thing runs through the synapses, then, with an even more honest face.
multiply by non-linearity (choice - like in a restaurant, according to taste)... and so on through
several layers, and the output is a beautiful neuron called FLET( FLAT, FLUT,
FLIT, FLOT-name you can choose to your liking) and if it activates, it means,
w-r-e-m-e-n-n-o-t.
What's so hard?
The quantitative estimation of market conditions on the selected TF can be mathematically precisely defined and expressed as a formula.
Let's assume that the market on the selected TF is considered trendy if the relative number of co-directional price movements exceeds 50%. Then a price chart that is directed in one direction (up or down) and does not have a single pullback will be defined as +1, and a price chart whose direction changes with every bar to the opposite and does not have a single directional movement will be defined as "zero". All other cases will take an intermediate position in the 0...1 range and will be characterised as "rather trending" or "rather flat".
Let's see how our sinusoid will look from this point of view (fig. on the left), if we find a trend factor for each TF, starting from TF=1 up to 25 (fig. on the right).
We can see that in this setting, for TF<8 the dynamics of this time series is trending (for TF=1 it is 100% trend), for 8<TF<23 the dynamics of BP is clearly flat (for TF=16 it is 100% flat).
In a similar manner, the market condition can be determined through the correlation coefficient between neighboring readings of the series of the first difference of the initial BP. For this purpose let's build the series of the first difference (y[i]=Open[i]-Open[i+1]) on the selected TF and find the correlation coefficient for each TF.
Then the price chart that is directed in one direction (up or down) and does not have a single pullback will be defined as +1, and the price chart whose direction changes with every bar to the opposite and does not have any co-rotation will be defined as -1. All other cases will be in the intermediate range -1...1 and will be described as "rather trending" or "rather flat".
The result is as expected, only the range of values taken by the trend coefficient has changed and the dynamics for a given BP has slightly changed. This is due to the fact that in the first case we only worked with the signs of the increments and in the second case we also worked with their absolute values. But it doesn't seem to be a matter of principle.
It is interesting to see what this function will show when applied to a real price chart, for example EURUSD 1m.
It may be noted that for signs of increments (fig. on the left), the market is more likely to rollback (<1/2) on all TFs up to one day (TF=1440 min). If amplitudes of increments are considered in the analysis, the market may be defined as a very weak trend (0<coef.<0.1).
It may be noted that for the signs of increments (fig. on the left), the market is more likely to rollback (<1/2) on all TFs up to a day (TF=1440 min). .If we consider the amplitudes of increments in the analysis, the market can be defined as a very weak trend (0<coef.<0.1).
This is a visual observation, but how it can be implemented in TS, what mathematical apparatus is better to use in this case?