1st and 2nd derivatives of the MACD - page 27
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
Briefly, in abstracts:
(1) this:
котри= тренд+ шум+ периодичность+ сезонность+ выбросы.
impossible to find, moreover, it is not true. there is no seasonality, no periodicity, no noise(!!!). This model does not work. Among other things, you will not be able to work with a quotient directly. I strongly recommend that you look for a transformation that brings the quotient to some stationarity
(2)
Everyone has focused on regression, with many posts with the usual nonsense, people don't know the terminology and don't take things personally that don't apply to you.
This is me trying to concentrate you on more important things. It is clear that not everyone is a mathematician etc.
(3)
You have to specify the stovepipe before you can evaluate anything. And this is TA, compared to which EViews is a huge step forward.
TA is not a "stovepipe", it's utter nonsense, though - colleagues who have been here a long time know my dislike of the thing.
(4)
The problem and the more important one - predictability.
By the way, how do you assess predictability?
This is to the question of fractality. As far back as 200 years ago, Hegel derived the law (one of) the transition of quantity into quality. This is when a system, having accumulated quantity, under a small action transitions to a new quality. This is now called a fractal.
And what does Hegel and fractals have to do with it? However, it does not matter. Let it be Hegel, for the first understanding will do.
If seriously, what are we forecasting? Qualitative jumps? News that will have an unknown effect on market movements?
The market model I gave above is a trend model, i.e. I'm going to use trends, Everything. I believe I either don't want to predict other things (e.g. vols) or can't (e.g. spikes).
Then prove the existence of a trend in the quotes, beforehand, remembering to give a clear definition of it.
I suggest we move to the specialised branch.
Moved my last posts there
The words approximation and interpolation are appropriate where there is a signal. DSP specialists keep forgetting that there is no signal as such in the marketplace
I cannot understand what "no signal" means. How is a digitized signal fundamentally different from an array close[n]?
I cannot understand what it means to have no signal. How is a digitized signal fundamentally different from a close[n] array?
Because a signal is a signal. For example a TV with a carrier and some other shit.
We do not have this signal and each of the kotir allocates what he wants, who is a mashka, who is a stochastic, etc.
The most important thing is that we don't have a goal to single out something like in DSP signal. We are highlighting something for future prediction
That's right, it's a regular tr-regression. I'm not so sure that these two sines and cosines will revolutionise the DSP, but try ticking off an article.
It's not very clear how you're going to adequately identify the model. I don't mean to firmly "inscribe" the model into a series, you can inscribe any model into any series with MNC (with some assumptions on accuracy). I'm asking about the understanding that the "optimal" parameters found will hold for a long time in the future, long enough to have time to work. There is a strong suspicion that the parameters will behave randomly.
Among other things, the model has a distinct disadvantage - you need to predict far ahead in order to profit from it. It is not very accurate, moreover, it does not describe the market at all, it will be seen by error analysis of the model - the first lags will be strongly correlated.
PS: although, there are a couple of thoughts about the development of this thing, if you are interested - I can write in private.
I described my old sine-wave model because someone asked for its output. I gave up on regression a long time ago and am not promoting my model here. For regression to be successful there must be a deterministic and predictable signal in the quotes that exceeds the noise. Natural phenomena are more or less predictable because of the repeatability of their physical assumptions (a famous example is predicting the number of sunspots). Such a deterministic signal in forex quotes, in my opinion, does not exist. There faa mentioned seasonality and economic cycles. Maybe on larger timeframes (years) they can still be spotted, though less frequently due to the influence of governments on the economy in their attempts to avoid a very fast growing economy with its inflation and the recessions which are the natural course of things. Imagine the task of predicting air temperature if people learned to regulate it with a certain precision in an attempt to avoid cold winters and hot years. If we are talking about quotes on small timeframes, which interests traders (minutes, hours), then there is no seasonality and cyclicity. There are only external perturbations (news) and market reactions. Since the direction of news cannot be predicted in my opinion, it makes no sense to include them in a market model (regression model, for example). The market reaction, i.e. the behaviour of the price after its initial surge up or down, can still be predicted, or at least I hope it can.
There is no such deterministic signal in forex quotes in my opinion
We can take a chart and plot the ZZ and we can see trends the day before yesterday, yesterday, today tomorrow and always. Just as physical processes are behind the weather, so behind these trends are economic processes, production processes, politics, etc. They are all inertial in nature and we can expect them to be deterministic trends rather than random wandering with drift. News cannot be predicted, but an individual news item very rarely has a strong impact on quotations, but a cumulative news item does. But in this case again there is some process behind it, which has manifested itself in the form of this news aggregate, which has influenced the quotes. Therefore, the basis of the trend is the inertia of economic and political life. This inertia is muddied by the emotions of the crowd, which we can see on the charts.
Faa mentioned seasonality and economic cycles. Maybe on larger timeframes (years) they can still be noticed.
If we are talking about your prognosis, then for me the following is of undoubted interest. There is no seasonal component in Forex. At least no one writes about it. But there is a periodicity, i.e. some wave with a variable period. In my thread C_4 gave a link to a very interesting book where the author proved that the amplitude of the wave with a logarithmic period grows before stock crashes, and begins to fade after the stock market crash. I.e. it is possible to predict stock market crashes! For example 87 97 etc.
For me this is extremely interesting in forex. No one has yet proven that there is no such periodicity in forex. I am not very interested in crashes, but being able to identify such a wave and increase the predictability of the model by taking such a wave into account is very interesting. The presence of such a wave can be seen by the distance between ZZ. It seems to me that the presence of such a wave has a significant impact on the concept of non-stationarity of the market. Maybe that is what we see in the variability of the dispersion.
There is no such deterministic signal in forex quotes in my opinion
You can take a chart and plot the ZZ and we will see trends the day before yesterday, yesterday, today, tomorrow and always. Just as physical processes are behind the weather, so behind these trends are economic processes, production processes, politics, etc. They are all inertial in nature and we can expect them to be deterministic trends rather than random wandering with drift...
The amplitude of a wave with a logarithmic period rises before stock market crashes and begins to fade after a stock market crash. That is, stock market crashes can be predicted! For example, '87, '97, etc... being able to identify such a wave and increase the predictability of the model by accounting for such a wave is very interesting. The presence of such a wave can be seen by the distances between the ZZs. It seems to me that the presence of such a wave has a significant impact on the concept of non-stationarity of the market. Maybe that is what we see in the variability of the dispersion.
First, if we take a perfectly random process with the same probability density function as the forex price and the same 1/f frequency response ("pink" noise) then we get a time series very similar to a quote. There will be "deterministic" trends due to the 1/f shape of the spectrum (the low frequency components are stronger than the high frequency components), spikes and high frequency "noise". Econometricians who do not know how this time series was generated will speculate about seasonality, cyclicality, trends, production inertia, etc. They will apply low-pass filters (mashups, the same Hodrick-Prescott) or ZigZag to isolate the "deterministic" components. But filtered noise will still be noise. Of course, the predictability of filtered noise is much higher as we know in advance that the filter output cannot change faster than our filter bandwidth allows. But the ability to extrapolate filtered noise has little to do with trade profitability.
Secondly, the observation of "logarithmic" waves before two crashes gives little information about their ability to predict crashes and even less information about their profitable use. How many of these waves were there in total and how many of them led to collapses? For example Robert Prechter of Elliott Wave International is famous for his "accurate" predictions of the bull markets of the 1980's, the 1987 crash, the bear markets of 2000-2002 and 2008-2009, and the March 2009 rally. But if you look at the history of his predictions, their accuracy is not that great. He predicted the 1982-2000 bull market would start in 1978, 4 years earlier. The 2000-2002 bear market came 5 years after his prediction (1995). And the 2008-2009 bear market should have come in 2003 according to his prediction. That Dow went from 8000 to 14000 over 2003-2007 is "within the margin of error of the prediction". According to his predictions, after a short rally in 2009, the market should have gone back to bearish and the Dow should have fallen to 2000. We know where the Dow is today. Prechter goes on to argue that his predictions are correct, it is the market behaviour that is wrong. They printed a lot of money, depreciated the dollar... He even plotted the Dow against gold where you can actually see that the market has been bearish from 2000 to today. How did that help his investors' profitability? Since 1980, Prechter's recommendations have resulted in profits lower than a simple buy-and-hold Dow stock. In short, "even a broken clock tells the right time twice a day".
First, if we take a perfectly random process with the same probability density as the forex price and the same 1/f frequency response ("pink" noise), we get a time series very similar to a quote. There will be "deterministic" trends due to the 1/f shape of the spectrum (the low frequency components are stronger than the high frequency components), spikes and high frequency "noise".
Lazy to look for the link, but attempts to distinguish a stochastic trend from a deterministic one are leading and there is some list of cases where this question is solved, but according to that publication this question is not solved in the general case.
The probability of encountering a stochastic vs. a deterministic trend is much more important here. We know that until recently the leading theory of markets was an efficient market with random walk. The Nobels are all here. But over the last 20-30 years this theory has been subjected to reasoned criticism and it has been shown that mostly deterministic trends operate in the market and almost no stochastic trends are present.
Secondly, the observation of "logarithmic" waves before two crashes gives little information about their ability to predict crashes and even less information about their profitable use.
I am not interested in predicting crashes, but I am very interested in the verbal model of the market. The success of forecasting depends on the completeness of taking into account the characteristics of the quotient: trend, noise, seasonality and presumably periodicity. I have listed almost all properties of a quotient for trend trading. The mathematician states that he sees one more property. If something has not been modelled, it means that something may be important. As in most indicators the noise is lost. Exactly in this sense I am interested in periodicity. On the charts we see it: price fluctuates around a trend or sideways line.
Correction: forex quotes behave like 1/f^2 noise (red noise). Here is the EURUSD M5 spectrum, 8192 bars (code attached):
The red line is a perfect 1/f^2 spectrum. The 1/f^2 spectrum is typical of Brownian motion. It is interesting to read this link on the subject of this thread:
https://www.mql4.com/go?http://includesoft.com/Statistics/About_the_Randomness_of_ the-_ FOREX_Rates.htm