Random Flow Theory and FOREX - page 24
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 still can't understand how you can work with an indicator that always shows one at the right end of the graph? What is its predictive potential - even if it is calculated according to a perfectly correct formula? I apologise if this is an idiotic question...
It's simple, I thought it was clear to you. The purpose of this indicator is not to predict, but to analyze BP (collection of statistics) by ACF type and parameters. Just imagine that after all our tricks the ACF view and parameters do not change. It means that it is a stationary process. We solve the inverse problem. We insert all of the obtained parameters into the process model and only then we will have a forecast and a fairly good one. If we manage to stay in the linear filtration theory, then Kalman's filter gives a priori a better forecast than any other filter (indicator, any mathematical tool). Mathematicians have proved it, and the proof is very rigorous and surprisingly even not a bit of common sense is not lost :-).
Neutron
Sorry to answer in the wrong thread, but the answer just really resonates with this thread.
Не согласен! по условию - приращения НЕзависимы. Любая локальная зависимось является случайной (стохастической), следовательно закончится так же неожиданно как и началась, а значит эксплуатировать это свойство не получится. Про второй вариант не понял. А вобще, попытка построить прибыльную ТС на случайном процессе (так как она определена выше) бред! Сергей, я же подчеркнул, что "нельзя в долгосрочной перспективе", и не исключаю вариантов локально выиграть. Это ничему не противоречит. Важно, что в среднем, на БОЛЬШОЙ истории доходность ТС (отношение общего профита к числу проведённых сделок n) стремится к нулю как 1/SQRT(n).
I think I understand it all, but my soul does not accept it, it's not right. I'll explain again with pictures, it's more convenient for me. For the sake of explanation, I have taken a long 1440 min. sample of GBPUSD, the latest data. To make it clearer, I built charts in MathCade and t=0 is on the left, unlike MT4 (on the right)
Fig.1 A chart showing how quotes for GBPUSD currency pair changed today, see Close[i] only.
Fig.2 Conversion of Yi (see Fig. 1) using the formula Сlose[i]-Close[i+1].
We have obtained the process of price increments and check if the increments are independent. Let's plot the ACF of this series of numbers as Fig. 3
Fig.3 ACF of the process shown in Fig. 2
What we get is that Delta is ACF, i.e. it is GSC and the increments are independent, there is no process correlation and therefore we cannot predict. But the strange thing is that with such a process of quotes (if they looked like that on Fig. 2) we may trade very well and profitably (if I am doing something wrong or have misunderstood, please correct). See Fig. 2 can=0 I stand at this point in any direction TakeProfit 10 points. During the day it is guaranteed to have about 5 trades. The only question is with the spread. With such quotes my brokerage company sets spread in 40 points. But in this case there is no market, as there is no point in trading.
Now let's return to the first picture, but the process is completely different in this one, even externally we can see (it is not Wiener's) that it differs from Fig. 2 and we get increments, as it depends on the correlation. Let's take a look at fig. 4
Fig.4 ACF of the process reflected in Fig.1 (x-axis normalized to the sample size)
What we can see is that there is a correlation time, i.e. during this time it is possible to predict; the second interesting point on the chart is a dotted line (B); it shows that there is an oscillatory process. Thus we get that in addition to the directed movement the blue line in Fig. 1 (it is y(x)=a*x+b plotted with MNC) is also an oscillatory process.
There is only one thing left to do :-),
- to pick up model(s) (in my opinion this ACF corresponds to an oscillatory link, as I wrote above on this thread).
- find out the lifetime of this model(s), if we have time to detect (identify) it before it collapses.
- and if we have time to build the TS.
Candid I have a request if not difficult check ACF Fig. 3, if the same, then acceleration check does not make sense and already BGS, if so the system of SRS will consist of 2 equations.
I have a request, if someone or met the literature where the type of ACF is determined by the type of process, tell me. Preferably with examples, which would be easier to understand and was more different models I have not enough of them.
1. You've created a (torn from life) TP and showed that you can earn on it. Well, you can build a profitable TS for such BP! So, what did you want to say? I guess that you wanted to bring this case as a counter-example to the question about the impossibility of creating a profitable TS for a normally distributed SV with zero MO... But, the question was about probability distribution of increments of such series and about the price series, which is obtained by integration of these increments, and you fulfilled the condition M=0 for the series itself. This is a point of principle. Take a series of increments for BP given in Fig. 2 and you will see a non-zero expected payoff with a great surprise. Therefore, construction of a profitable TS for this realization is possible and there is no contradiction.
It seems to be simple.
The ACF by definition shows the relationship between different BP values. How is the ACF shown in Figure 4 constructed?
Candid I have a request if it is not difficult to check ACF Fig.3, if the same, then there is no point in checking acceleration, if so the SRS system will consist of 2 equations.
Secondly I think it 's just a bad signal to noise ratio. If we generalize raw data to Close[i]-Close[i+m] we get this:
The lower curve corresponds to m=1, the middle one to m=60. Thus, we can conclude that the market is merciful: those who want to get rid of noise will receive it, while those who want to get rid of noise will receive ... Momentum :). However, as they say, you don't walk in someone else's garden with your own stones :)
The ACF by definition shows the relationship between different BP values. How is the ACF shown in Figure 4 constructed?
But we meant probability distribution of increments of such series and price series, which is obtained by integration of these increments and you fulfill the condition M=0 for the series. This is a point of principle. Take the incremental series for BP in Fig. 2 and you will be surprised to see a non-zero expectation.
If I understand correctly, we should take Close[i]+Close[i+1]. Then we get a series of numbers of Fig.1 with accuracy to const. But my statement that the process of Fig.1 is not Wienerian does not seem to cancel. The ACF view says otherwise. If I'm wrong, let's go to the other side and prove that it is Wienerian.
And here is an example a little more interesting, with this strategy I will earn at any kind of law distribution of std.magnitude need to fulfill only one condition can and sko = sont. So it raprostranlyaetsya not only to EO with can=0.
Candid I have a request if not difficult check ACF Fig.3, if the same, then acceleration check does not make sense and already BGS, if so the system of SRS will consist of 2 equations.
Second, I think that returns is just a bad signal to noise ratio. If you generalize the raw data to Close[i]-Close[i+m] you get this:
The lower curve corresponds to m=1, the middle one to m=60. Thus, we can conclude that the market is merciful: those who want to get rid of noise will receive it, while those who want to get rid of noise will receive ... Momentum :). However, as they say, you don't walk in someone else's garden with your own stones :)
1. I tried and Y-mu, seems to simply reduce variance (although I did not exactly check). I wanted to once again make sure in the thesis of inability to earn on forex in the long term, but it is from another branch.
2. The signal/noise ratio is bad, I think. But I always wonder what is signal and what is noise. What do you mean by that, because when you build a model there is also a notion of excitation noise (EIR) and in clever books they write with Mg=0 and 1 intensity, in this intensity and the dog is buried, because it's not sko and not dispersion, and in general often has such dimension mama dear. So while I do not have a composter (went skiing for 2 weeks) sits here all sorts of nonsense I think yes books read. In general, I understand everything like a dog, but I can not say, and I can not tell the stupid machine (computer) can not explain what I need from it :-)
"Well, as they say, don't look for stones in someone else's garden :)" Well, your stone is exactly where it's needed. Even if you start throwing them at me, I will try to understand why.
Yeah, and on the subject of ACF not getting on the screen, try rationing it like I did in the 4th picture. It turns out constant and at outputting "interesting points" you may just take it into account. But I don't know if it is possible to do it in MT4, in matcad by easy hand movement :-).
Edit. Almost forgot to ask Momentum - is it an installation? i.e. there is also the concept of mass? If so, how is it calculated? Very interesting bouquet may turn out, force is there, energy is there, inertia too it seems, there is mass left.
There's no such thing as a Wiener process. The Wiener process is an integral of Gaussian noise. But the returns process does not look like Gaussian noise. Fig. 2 demonstrates that it does not even look like stationary (on minutes; on ticks the process looks completely different).
From here a bit more details, what does it mean completely different ? I think you should have a picture of something similar to Fig. 2 only for ticks, post it with explanations ? Maybe you really can't even use minutes, I just haven't got a clear answer for myself yet.
Nah, Prival, Momentum is an indulgence that is calculated by the simplest obviously non-physical formula. Here is the link: https://www.metatrader5.com/ru/terminal/help/indicators/oscillators/momentum. There is also ROC, something similar: https: //www.mql5.com/ru/code/9340 .
About ticks - here's a link to a thread with my attempts at tick research: 'Tics: amplitude and delay distributions', see my last picture on the first page of the thread (tick process for oira). 99.5% of all ticks are +-1 and the rest is unaffected.