Probability assessment is purely mathematical - 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
"Like how to trade using ACF?" ACF is a market analysis tool and cannot be used to trade (or rather I have no idea how to do it). ACF allows you to visually see and calculate the parameters of the processes occurring at the moment in the market, but you need to understand what ACF is and what it is all about. Try without a computer, a piece of paper and a pen to build the ACF and much will become clear...
You shouldn't do that.
You can trade.
Just have to decide which trading horizon is preferable.
For medium-term trading it is perfect. Well, no one is looking at the long term here.
Have you sorted out the 40 and 20 periods?
Or do you keep insisting that you should expect "miracles" with the original and the shortened one by removing every other observation?
I'm not discussing "miracles" at all. I've already written on the subject. Once again: I believe that the pictures I cited prove that the same BP (EURUSD in my case) on different timeframes (M30 and H1 in my case) has different statistics. If you are interested in this particular assertion, then I am willing to discuss it, Krylov does not want to discuss it.
"Miracles" I am not discussing at all. I have already written on this subject. Once again: I believe that the pictures I cited prove that the same BP (EURUSD in my case) on different timeframes (M30 and H1 in my case) has different statistics. If you are interested in this particular assertion, then I'm willing to discuss it, I don't want to discuss Krylov.
1. .. I know what ARPSS are, and I gave up on them a long time ago, and regression models too (I think you have the same conclusion).
I don't have that conclusion, but the idea of using that package somehow. On this forum once there was a man who claimed to have managed to use the package to build a profitable strategy on the real.
According to feedback on the package, it requires a high level of skill and attention from the user, despite its outward simplicity. The main thing is that it is very labor-intensive to obtain significant results.
2. Your charts are strange.
I cannot say anything about the last graph yet.
By definition ACF at 0, equals 1 (for any data) and gradually decreases to 0 (when the bias is completely out of sample). I have not seen this in any of your graphs.
Regarding the first two ACF plots, I would like to make a crucial point for this forum.
It would seem that ACF of the first candle should be equal to 1. But the developers have never forgotten that they are dealing with a random value. That's why any statistics must be necessarily accompanied by at least two values: deviation and confidence in this deviation which is what is shown on the charts: the ACF value is given and the RMS to the left.
3. These are the phrases that baffle "...Let's look at the Close chart without ACF. It looks more like noise. But we are trying to get rid of the regular component..."
Also, by trend I mean the straight line equation y=a*x+b. That's what I subtract from Close before plotting ACF. And after that I calculate ACF - and its form clearly shows that in addition to the trend (straight line equation) there is always an oscillatory process in the market.
I do not know how to explain.
1. We take Close.
2. Subtract the trend (y=a*x+b) from the Close.
3. Construct the ACF.
Each Close plot shows what was subtracted from that value for detrending. I may have named the graph incorrectly, but the graph itself shows analytically what was subtracted. This eliminates the ambiguity of interpretation that I allowed.
From what I've learned about the package, the fundamental thing about its success is the preliminary data preparation that precedes the model identification step. You take the simplest equation for a trend. On what basis? I gave two different equations for detrending (the regression is more complex, obtained by the package) and calculated (by eye) that it is better if the ACF is subtracted from the Clause. Much of the package contains tools for detrending and removing the cyclic component. I think the difficulty of using it is precisely this point (apart from the limitations inherent in the idea of the package - the stationarity of BP). What you cook is what you get.
4. We get the following picture https://www.mql5.com/ru/code/8295 red is what we get. At that particular moment in time it is the oscillatory 2nd order. There is also ACF plot obtained from the model with correspondingly calculated coefficients (blue plot), practically a perfect coincidence... But it is not always the case, today this ACF may look different, respectively a different model operates in the market...
I am not ready to discuss this yet.
isn't it obvious to you that period 40 will become 20 on a shortened row?
Jesus, what does the period have to do with it if I'm bringing in a spectrum analyser that uses the Parks-McGlellan algorithm to build filters.
isn't it obvious to you that period 40 will become 20 on a shortened row?
I think that's enough of an enlightenment. The man is competent enough. Let him figure it out for himself. It is a pity that he believes in packages and programmes that they will do everything for him. He needs to understand the essence of these transformations.
Faa1947 13.09.2010 19:01.
If you do not believe that the traders at this forum are competent enough? and all these transformations have been chewed over here more than once... you should at least read carefully what you attached as an archive to the post... there is a link to this forum at the end.
If may be you do not allow an idea that this forum is full of competent traders and all these constructions have been chewed over here more than once... you should have read carefully what you attached as an archive to the post. there's a link to this forum at the end.
I can't judge the quality of traders - I don't have their financial reports.
I assure you that I have been following the forum for a long time, I have been participating for some time. I have a low opinion of this forum. Apart from the obvious drawback - widespread participation by ignoramuses and reinventing little bikes, the main drawback is the overwhelmingly fragmented nature of the topics discussed, including the thread on your link.
I am not promoting STATISTICS, besides it there are a large number of packages often with greater capabilities, e.g. Matlab. I am advocating the use of any package rather than reinventing the wheel. By using perhaps the most primitive one - STATISTICS, you can avoid many small mistakes and always solve the whole problem from analysis to forecasting. There will be other problems caused by the package itself, but it's less of an evil than reasoning about the basics of statistics and theorists.
I think it's an interesting idea and I'll implement it (I'll try), but at another time and in another thread.
Thanks, you, Prival, for your opinion on my posts.
...
I will try to help FreeLance, he wanted you to understand the simple truth that if you feed different data to the spectrum analyser, you will get different spectra. It is very simple and clear. And it's not proof.
your quote
"This is not the first time I see your opinion about ticks. To my mind ticks statistics has nothing to do with timeframes statistics, and each timeframe has its own statistics, and one cannot be deducted from another. It is deductible at the level of analytical indicators, not statistics.
As a proof I propose two pictures. EURUSD30 has 7200 bars on one of them. On the other one, EURUSD60 is 3600 bars. We have different Fourier decompositions!"
You input different data, and that's why you get different images. And it turns out that it's not the Fourier transform but something else (it turned out only after several pages).
And concerning the ticks using candlesticks (bars) you are extracting from the tick stream, in statistics it's called censoring. And the worst thing is that you do not control this censoring during analysis. This is what makes ticks valuable, but you can't get ticks from bars; you may try to do it, but you lose much information, you lose the structure of the tick flow, its intensity, density, temporal distribution of ticks... Here you can read my another battle with developers https://www.mql5.com/ru/forum/1031
So farI have reached my opinion, the history in the form of a tick chart is better than the history in the form of bars - better from the viewpoint of market analysis (research).
Oh I forgot, that's just for you, for education. Input this BGS algorithm, not EURUSD30 - 7200 bars, but 7200 BGS counts and see what happens. I think the results will surprise you...
I will try to help FreeLance, he wanted you to understand a simple truth: if you feed different data to the spectrum analyser, you will get different spectra. It is very simple and understandable.
That's exactly what I'm trying to say. The quotes on different timeframes are statistically different.
And concerning the ticks, using candlesticks (bars) you are extracting from the tick stream; in statistics it's called censoring. And the worst thing is that you do not control this censoring during analysis.
Why control it? The quotient itself of the corresponding timeframe is of self-sufficient value.
... here you can read my next battle with the developers https://www.mql5.com/ru/forum/1031
I've read it carefully, but the questions remain.
This is what values ticks are, but bars cannot be used to obtain ticks, you may try to model them, but much information is lost, the structure of the tick flow, its intensity, density, ticks time positioning is lost.
When we measure the temperature of the body, much, perhaps, valuable information about Brownian motion is lost. So what? We are interested in temperature, not the molecular mechanism of its formation.
Each timeframe reveals some new information about the market. And from a small timeframe chart I cannot get even important information: what may be considered as a trend on a small timeframe may look like a correction or a reversal on a higher one (in this case it is a trend on a lower timeframe).
When you are writing about losing information when leaving the ticks, it is unclear, what for is this information? How can it be used for predicting a quote and on what time horizon? Is it possible to predict the next M5, M30, H1 candle, etc. from the ticks? And is it possible at all?
I have doubts that it is possible to obtain the answer to these practically important questions for the following reason.
When analyzing ticks, they ignore a very important circumstance: all ticks are different: one tick has the volume of 0.1 lots, another tick has several hundreds of lots. One tick is from a pipser, another tick is from an investor, and the third tick is from a hedger. Each works on its own time horizon. Whereas the stock market has known volumes of positions, we have no volumes. But of all types of financial markets, there is no information about the time horizon for which the pose arose. I remember now that there is a trading strategy: track the buying volumes of large funds and buy as well.
I can't entirely agree that the construction of a senior timeframe is obtained by censoring. Usually, BPs are censored which have some measurements corresponding to important events: took medication and then died. All intermediate dimensions can be skipped. Of course it would be very interesting to censor a cotier built on tics with a volume greater than (less than) some value. But it is not that important.
Judging from the literature, the enlargement of the timeframe allows to clear the quotire from the noises (pips) and to focus on the larger players, which determine the market movement.
The incalculable number of factors piling on top of each other, pushing something in this world, stabilises and simplifies it out of all proportion. ............
I doubt you built a model with "innumerable factors" or anything like that. More likely you trusted your intuition. But imagine that there are models that explain price behaviour. And for example, to argue that within the above model price is absolutely random, you would need to show that the difference in sentiment is absolutely random, that yesterday's prevailing desire to buy has no effect on today's. To do this, with a bit of work and synchronising the model with the real market, you will get a confirmation answer - the sentiment is absolutely random. Or you won't.
It's not as simple as it sounds. Elioters or moving averages are just an add-on to price. It's easy to accuse them of self-delusion as they juggle the consequence. Try to address the cause, to look beneath the price, so to speak.
Let's look under the price .......
A model explaining the behavior of the price, that's exactly what you Vita is, a rather primitive and prone to doubt model, or even a template that takes common stereotypes as tendencies, to explain and nature of which you substitute the values that you have managed to see in them at the moment! It is comfortable for you, it is understandable to you and it is normal! For you! (please excuse the model)
Tomorrow (the future) will come, regardless of what you want, and much faster than you can imagine. And it is inevitable. The whole point is that none of us knows what that future will look like. At that moment the stereotypes stop working, the market does anything but what you expect and at that moment the tentative stability turns into armageddon.
The nullity of those tools and tricks, with which you are planning to make a profit from the trends that you understand, that grow into trends, that are based on yesterday's sentiment and have clear patterns and consequences, is nothing more than a modeling of a situation. And the situation has nothing to do with what is actually happening in the market! I'll tell you a secret - no one knows that! But everyone makes forecasts, which with a probability of 50/50 turn into legends (about 5% of successful traders), and then into apocalyptic nightmares and promissory notes.
The conditional stability that you have traced back to history generates a desire to act in accordance with the current trend and that is what the vast majority do. Thereby creating a so-called market memory. Stereotypes breed stereotypes and that's true. But you can't make money from it. This is not what you need, unless you are writing another masterpiece on how I have learned the tricks of the trade.
I have offered a model of behavior in the market completely devoid of convention and prejudice, because I have simply excluded the causes and effects, those with which you so easily manipulate the situation creating fictitious market patterns.
There is no point in explaining something which has no practical application. There is no point in wishful thinking. There is no point in wasting time on something that defies all rules and stereotypes.
I am talking about what will happen tomorrow or in 10 minutes.