Random Flow Theory and FOREX - page 29
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
By the way, just so you don't miss it, Better uses a two threshold decision logic, from my point of view it is also the most efficient, I used to draw it in pictures before.
Here's what he told me
In fact, the output is three options:
It's good that not everything I've written here is bad and doesn't work. There are still some sensible thoughts in this stupid head. The only thing I would add is that thresholds (A) should not necessarily be equal, but should be recalculated all the time depending on input statistics. At least, I would try to do so. Eh, I wish I could get to this place of research. And now all stupor without checking the adequacy of the model to move forward meaningless. If I can't calculate NUT then there's nowhere to shoot :( books are needed which are clever.
By the way, just so you don't miss it, Better uses a two threshold decision logic, from my point of view it is also the most efficient, I used to draw it in pictures before.
Here's what he told me
In fact, there are three options at the output:
It's good that not everything I've written here is bad and doesn't work. There are still some sensible thoughts in this stupid head. The only thing I would add is that thresholds (A) should not necessarily be equal, but should be recalculated all the time depending on input statistics. At least, I would try to do so. Eh, I wish I could get to this place of research. And now all stupor without checking the adequacy of the model to move forward meaningless. If I can not calculate the NUT, then all do not shoot :( books need a clever.
I would use slightly different logic.
Short
Close Short
Long
Close Long
Do nothing.
Close all
I see it differently (All these four actions Short, Close Short,Long,Close Long). This should not be the output of the recognition algorithm, it can be NS. These are actions of the trading system, a consequence of what is given to it as an input (to shoot or not to shoot). And it has to do it correctly (at the right moments). The TS should be given to it for entering
I.e. with respect to the trajectory being analysed, many alternative hypotheses are put forward for its possible motion. This is what I mean by the classes that need to be recognised, and there is a zone of ignorance between these classes of motion. I.e. until one of the hypotheses is chosen by some criterion, I don't know what to do. And when the decision is made (the enemy's behaviour is recognised). The algorithm for analysing when to shoot where to shoot and whether it is necessary to do it at all now.
So it is like this.
After all, Better also does the same at the output HC up or down. And if you take Short, Close Short,Long, Close Long as an output, it is difficult Reshetov, I hope you do not need to tell how perfect in this case, the algorithm. I certainly will not go into battle with it (Reshetov's algorithm or any of its modifications).
As promised, I am posting the synthetics (price flow model), just add the straight line equation.
Methodology
I've written before, but I'll repeat, it is necessary to dissect the process into its components. Achieve that in the residuals would be BGS.
1. Subtract trend y(x)=a*x+b
2. Construct ACF, by the parameters of ACF determine the parameters of oscillation, insert them into the equation of the oscillating link.
3. Subtract. Check residuals after subtraction check at BCS.
Repeat steps 2 and 3 until acceptance of hypothesis that the residuals are BGS, say, according to Neumann-Pearson criterion with error level of the first kind 0. 05.
Then, we write down all components of the process as matrix F (see page 1) and simulate the process.
We get something like this.
And everything seems to be true, everything is similar, here is the solution, but ..... is very good, just great. We have to start all over again. The data on which the model is based is not accurate. You need a model as Candid said of that continuous world quoting process. I now understand where the "tick lags", gaps, all that non-stationarity in general come from.
And I was wrong in thinking that it's not a number that rules the world, but a function (as I wrote about earlier).
The world is ruled by a NUMBER, by knowing that NUMBER you will know the world, and by controlling that number you will control the world !!!
And YOU know what this number is called, you have heard its name dozens of times. I won't tell you its name now, but I will later, because it's worth YOU thinking about what I'm talking about.
Interesting conclusions I get, any representation (transformation) of a stream and its reflection in the form of bars is wrong. And the most surprising is that ticks are not precise, they should be transformed, but we should think about it, it will look better in the morning.
It would be interesting to hear your variants, maybe you know this number, I was just not told about it. And I'm discovering America all over again.
to Prival
P.S. 2 Prival: I'm reading Amir's article ( 'The principle of substitution of time in intraday trading' ), trying to please your inflamed imagination looking for physical analogies. See:
From the point of view of statistics of random processes, it has long been accepted that the process of price change in the first approximation is some kind of diffusion, that is, the process of transfer of matter or energy from an area with a high concentration to an area with a low concentration. Maybe, it is better to describe this process not in radar terms, but in diffusion terms? There is a game on the difference of bets (I think it is called carry trade). Here you have different concentrations and natural energy transfer(carry = carry)...
This is indeed a very good idea. I just use in my model a similar approach that I wrote about briefly some time ago, namely about "flow" of energy between linear regression channels (of course, it is not necessary to use LR, but it is easier to count it). I recommend to remember attractors, or rather pseudoattractors, may come in handy :o)
Have you tried this? I think I already gave you a link to this programme - http://www.r-project.org/
Not addressed to me, but tried to use the link. Rosh, I'm ashamed to say, but it didn't install. Please tell me where the problem is. Help says it's easy to double-click the icon: To install use `R-2.6.1-win32.exe'. Just double-click on the icon and follow the instructions. But this icon is nowhere to be seen, and in the downloaded folder of executable files not found at all. Where am I stupid?