Market patterns - page 7
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
I have studied the angle and stages before, but did not get used to it / did not get used to it/.
I've studied the angle and the stages before, but they didn't stick / didn't work out / but in general I thank you for the "reminder" of the angle of inclination of the price series, I will interpret it...
And when the Bid-Ask spread widens in both directions what is it: a flat, a "double-sided" trend or a trend?
And when the Bid-Ask spread widens in both directions, is it a flat, a "two-way" trend or a trend?
You want a universal formula that explains everything from milliseconds to years? I'm sorry, I don't have one.
I analyse from M5 through H1 to D1. What kind of spread expansions can we talk about in this context?
Spread extensions are HFT algorithms, trends are trends lasting much longer than a day.
We are talking about incomparable things. I wrote above that even on M5 it makes no sense to look for trends.
No, I am just really interested in how people who rely on notions such as flat / trend explain such things for themselves.
Another example, trend/float/trend during the trading session and similar behaviour at night/holidays/weekends. To what extent are these time periods equivalent? I'm talking about traded volume. Where is even a hint of this in the definitions/notes given?
It's you who then(I address figuratively ) "unreasonably" exclude the "wrong" primes from your analysis, intuitively correcting the "basic" concepts.
pantural : If it's trending ... If flat then ... Two opposite assumptions depending on the type of state. Recognition of states is difficult (many methods) and the main puzzle is in its recognition, or rather in how to unravel it.
Here we are.
Baba Klava at the wharf, quietly under the apron exchanging dollars for hryvnia and hryvnia for dollars, constantly wondered how she managed so well to recognize the state.
Yes ... and she's got more income than most of the community here. And you can't even argue with that, market maker.
The trend (trend) is the suppressor (what suppresses) the bull or bear party (as the case may be), the mental component of the market that is in the heads of most market participants.
The suppression of one part of the market leads to a flow of participants from one conditional party to another, resulting in a quadratic relative increase of participants in the winning party. Hence the nature of momentum.
Since the component is mental, it changes in accordance with the psychological laws of nucleation, development, and conquest of life space (minds) by ideas.
At the lower TF it is very dynamic, changing almost without inertia, while at the higher ones there is inertia (which creates angles of movement).
The trend has only one characteristic "angle". It is also a force of pressure. If the angle changes we can talk about the degeneration of the trend or its completion.
Volatility is not related to the trend characteristics (it does not define its development), but it is needed for making a profit. Therefore, it would be reasonable to measure it in TS and use it as a reference point.
Trends have 3 stages:
The first occurs in the lava of opposition, and is characteristic of the 3rd stage of the ruling trend. The second stage is full control of the ruling party, with inter-establishment wars within.
The development of internal confrontation (when some want to go fast and others are in no hurry) leads to a transition to the third stage. Thus, the emergence of a new trend takes place in parallel with the withering away of the old trend (this is the stage of uncertainty, the decision made in this stage is the most expensive, the criticality of error is high), work in the 2nd stage is jumping on a moving train (often ineffective, but with the right skill can be drunk without much risk).
I thank Uran for a less clear definition of a trend. Although the concept of "angle" is rather relative, it is better to use the term "increment" as "angle" depends on 2d interpretation and arbitrary scaling.
Trend(pronounced"trand", fromEng. Trends can be described by various equations - linear, logarithmic, power-law, etc. The actual type of trend is determined on the basis of its functional model selection by statistical methods or smoothing of the originaltime series.
Wiki.
IMHO The easiest way to find a trend is to calculate the increment from the price smoothed by MA. The sign shows the direction and the modulus the intensity. If the modulus is less than a certain value then it is a flat. By "price" we mean a value containing maximum statistical information of a given FI, for example OHLC/4. As for the structure ofOHLC intrinsic distribution, let's leave it for later and suppose a small influence of this data for the time being . This method is in my opinion the most "logically clean".
TF=TD((k*(SF(price(t),n)-SF(price(t-m),n)),x)
TF-trend/flet(1,0), k- scalefactor , SF- smoothing filter, price-price, t-time, n-period SF, m-step differentiation, x-value of threshold trend\flet, TD- (threshold) threshold function.
SF is as a rule any MA (SMA,EMA etc) for retrospective analysis smoothing filter(s) is bilateral, i.e. without lag, and in trading it is selected according to the data, left-side filter.
Flat in such definition is really a kind of trend. But it is not clearhow to select coefficients (k,n,m,x) and SF to give a correct signal for changing the TS type from trend following one to channel one.
It is believed that this is achieved by enumerating parameters in testing, full or any accelerated (genetics, montekarlo, etc.) but I'm tormented by vague doubts why that is ...
The results are very strange, at least for me. I feel as if I'm walking blindly, or in a scuba tank committing copulation...
It's like the main idea is there, but I can't grasp it yet, I can't even formulate the question clearly. In general, gentlemen who reproach me for not being specific, I am entitled to it.
But as they say, the worst question not asked.
It is commonly believed that it is achieved by testing, full or accelerated (genetics, montecarlo, etc.), but I have my doubts...
Without testing there is no way, all the power is in the statistics. In general, it is the only data which can be trusted. The trend or flat is not important, what matters is the pattern that anticipates statistically, what is the potential for price shifts. In general, we probably can not introduce specific market conditions with names and associated rituals. Just average the patterns followed in hindsight and then compare the current one with such generalised benchmarks. I think at each scale and instrument, there will be a different pattern.
....... In principle, it is probably possible not to introduce specific market states at all, with names and associated rituals. Just average the patterns that were followed in hindsight and then compare the current one to such generalised benchmarks. I think on each scale and instrument, there will be different patterns.
That's one direction. But not a panacea. And the direction is the most ornate. You can implement such recognizers on mql5 according to a ready-made plan, but the development, testing and training algorithms, is easier in a set of specialized DataMining environments and/or inhouse developments.
Believe me, the tempting brand of "neural networks" in the hands of a dilettante is no better than the MACD+martin tandem, just by trying parameters and standard grid architectures, nothing will work. This is the most complex fusion of science and art.
You can't just mix and match all the patterns that precede a desired reversal or trend, just to average arrays of prices... that's nonsense. The patterns themselves should be represented rather intricately, by several levels of "derivatives" of different orders, representing the most essential part of pattern information. Such coding itself is alchemy, which, though it has some basis in itself, but still it is refined and undergoes selection and the result is completely logically difficult to substantiate in all details, cluster, correctly describe atalons for each class, etc. And then neuronki is the highest wizardry. But I must admit they are unmatched so far, with the right approach.
I advise to deal with one neuron and learning network with it, then two, one layer, two, etc. by visualizing what distributions of weights appear in learning which data by which learning algorithm, starting with simple series of what periodic functions, then more complicated patterns, etc. We should try to see how the grid thinks and learns. I should not try to guess what will come out just by searching for parameters and filtering the initial series.
Anyway, I get it! It's a great illusion!
Yesterday I had a discussion with a professor of mathematics, who was defending his thesis on a very non-trivial infinite dimensional functional analysis, although I have technical degree, but I did not understand what his thesis was about even when trying to explain it "on fingers", or rather on fingers it is inexplicable, just like it is impossible to imagine that in >3d dimensions for example you cannot tie a knot. Very tough dude.
Well, he proved to me on different examples that the surcharge on the effective component, which sort of sometimes appears in all sorts of algorithms searching for non-random connections, exists only in hindsight, but in reality is entirely eaten up by insiders, that is those who have more (qualitative) information compared with the crowd. That leaves only the SB. Most of the obvious and unobvious correlations are already taken into account in spreads and commissions at the inter-bank level, and the brokers, those of them who output to the inter-bank, exaggerate this accounting by 2-5 times. And there is a majority who don't know this and a minority who do.
So, another impulse of my interest in getting rich quick has crashed on the granite of science.
I get it! It's a great illusion!
It turns out that non-Insiders are only allowed to have a near-market income. This, of course, is not so beautiful in my view in terms of profitability. Although I am a seller of illusions by current profession, I don't sell knowingly dangerous things, I just push pictures.
But in the market niche, it is not, if you take as axiomatic qualified opinion my friend scientist, it turns out that from a social situation in which I can only pennies earn selling attractive to test(past data) automatic systems, through sites, but it is 3 (ue) figures a month, and so I earn without kidnapping and stoka.
If you consider scamming, it is only in the context of super-profits. And this cannot even be called "profit", it is more like a qualified office plankton's salary.