We need a second head, or even two, like the Garrynych kite. - page 11
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
.......
You seem to have some knowledge of the ACS.
Then imagine a tank connected by two gates to a pipeline through which the fluid flows turbulently.
The task is to fill the tank with liquid by alternately opening one of the gate valves (buy or sell), depending on the direction of the liquid.
The delay in opening and closing the gate valves (spread) must be taken into account. If the wrong gate valve is opened, the liquid will flow out of the tank (wrong opening direction). If two gates will be opened (with equal amount of liquid flowing out from both gates at the same opening, i.e. double the spread loss), the volume of liquid in the vessel does not change (locke, forgive me the reasonable part of the community for this vile term, :) ). There is one more problem - our hypothetical ACS has possibility to take measurements only at equal time intervals (discreteness). For all that, the nature of fluid movement changes over time.
MA say? Not suitable (Why? sorry, I won't repeat, it has already been said).
One of the software packages, capable to simulate this problem (I mean, of course, one of the most convenient for our problem, though surely it is possible to put similar experiments in matcad), is TraceMode. The fluid flow will be the imported quotes.
I have not tried it myself, I have not got round to it. Dare, the results are sure to boast. :)
You seem to have some knowledge of the ACS.
Then imagine a tank connected by two gate valves to a pipeline through which fluid is flowing turbulently.
The task is to fill the tank with liquid by alternately opening one of the gate valves (buy or sell), depending on the direction of the liquid.
The delay in opening and closing the gate valves (spread) must be taken into account. If the wrong gate valve is opened, the liquid will flow out of the tank (wrong opening direction). If two gates will be opened (with equal amount of liquid flowing out from both gates at the same opening, i.e. double the spread loss), the volume of liquid in the vessel does not change (locke, forgive me the reasonable part of the community for this vile term, :) ). There is one more problem - our hypothetical ACS has possibility to take measurements only at equal time intervals (discreteness). For all that, the nature of fluid movement changes over time.
MA say? Not suitable (Why? sorry, I won't repeat, it has already been said).
One of the software packages, capable to simulate this problem (I mean, of course, one of the most convenient for our problem, though surely it is possible to put similar experiments in matcad), is TraceMode. The fluid flow will be the imported quotes.
I have not tried it myself, I have not got round to it. Dare, the results are sure to boast. :)
I presented the tank, but I can't replace the liquid with quotes, but otherwise it's OK, the virtual simulation process is going on! However, I don't know how to attach TraceMode here, as it seems to be oriented towards production processes.
I didn't claim that Mashki is perfect and solves everything, and the topic wasn't created to chew over an asbestos problem everyone's been experiencing again.
I have been struggling with the adaptation of TS to the market for years, I have reinvented a bunch of TS, trying to find the levers (your terms - valves), moving which allows in dynamic mode tuning TS parameters adequately to market changes. So far I identified one pattern, the more multi-parameter TS is, the more flexible it is and the higher its performance can be tuned, but, in proportion to this deteriorates its stability and narrows the operating range. And the more tunable parameters there are, the more complex is the algorithm of their interrelation and the more difficult is to bring the whole system into the equilibrium state, as well as to formalize the dependence of these parameters on various market phases.
And Mashka is chosen as an example that everyone can imagine, everyone has more or less experience with them. Furthermore, according to my experiments, their potential is far from exhausted. If we do a system based on the same flip-flops with feedback, it can significantly improve not only the amplitude characteristics, but also the phase response, i.e. reduce the lag of MA, although, of course, you will not get a super system. If you want to experiment with it, apply not a negative feedback, as in an amplifier, it improves amplitude characteristics, but the lag increases even more, and the positive feedback, as in generators, it leads the system to an excited state, but imposing boundary conditions can keep it in equilibrium, and due to PIC can improve not only the amplitude but also the phase characteristics.
I have left these experiments for the time being; I have not got to a finished TS. I am currently developing an indicator specialized for manual trading and visual testing of strategies. I think it is not very rational, in order to check the strategy it is necessary to make TS, and although it mainly comes from ready-made modules, nevertheless it requires time for programming and debugging. And this indicator will allow in a visual mode in the tester to simulate manual trading at different phases of quotes history, and immediately assess whether the strategy is worth attention or not. (If anyone does not know how to trade manually in the tester, to speed up manual testing as compared to demo account: I manage orders in my EA using global variables, I start the EA in the visualization mode, attach the needed indicator to the window, set the suitable speed of running, as soon as the visual combination of signal lines is formed to open a position - I press pause, F3, by calling global variables window, correct the value of global variable for the corresponding position in the window, close the window, shoot the cursor.
It seems to be visually similar market phases and indicators readings of signal lines, but there is some imperceptible shift and logic of the TS works differently. In fact, I was counting on the brainstorming and set this task, but it seems it is not interesting for very few people.
And in general, it is time to go on holiday, to the sea, in the sun, because I'm tired here, training bulls and bears.
Don't waste your time - you won't make any money by randomly wandering around. If you keep stubborn and stubborn in this direction, in a year, two... you'll be writing the same thing here... There have long been special tests that can be used to assess in advance whether you can make money on trading or not.
I don't quite understand your statement. Do you mean that technical analysis is not working in principle and its use is useless, and therefore the adaptation of TS to market changes is not possible? Because I'm not saying that I cannot make profit, but that the system of feedbacks that adequately tunes TS according to ongoing changes cannot be simulated.
And what special tests are you talking about, could you give us some examples?
1) The tank has presented, the liquid on the quotes as it turns out not to replace, but otherwise all is normal, the virtual process of modeling went! True, I do not know how to attach TraceMode here, I think it is oriented production processes.
2) I didn't claim that Mashki is perfect and solves everything, and the topic was not created to re-chew the problem that everyone is sick of.
3) I am struggling with the problem of adaptation of TS to the market for many years, I have changed a bunch of TS, trying to find the levers (for your terminology valves), moving which can dynamically rebuild the TS parameters adequately to market changes. So far I identified one pattern, the more multi-parameter TS is, the more flexible it is and the higher its performance can be tuned, but, in proportion to this deteriorates its stability and narrows the operating range. And the more tunable parameters there are, the more complicated is the algorithm of their interrelation and the more difficult is to bring the whole system into the equilibrium state, as well as to formalize the dependence of these parameters on various market phases.
4) And Mashka is chosen as an example that everyone can imagine, everyone has more or less experience with them. In addition, according to my experiments, their potential is far from exhausted. If we do a system based on the same flip-flops with feedback, it can significantly improve not only the amplitude characteristics, but also the phase response, i.e. reduce the lag of MA, although of course, you will not get a super system. If you want to experiment with it, here's a tip: apply not a negative feedback, as in an amplifier, it improves amplitude response, but the lag increases even more, and the positive feedback, as in generators, it leads the system to an excited state, but imposing boundary conditions can keep it in equilibrium, and by PIC can improve not only the amplitude but also the phase response.
5) I have left these experiments for the time being; I have not got to the finished TS. I am currently developing an indicator specialized for manual trading and visual testing of strategies. I think it is not very rational, in order to check the strategy it is necessary to make TS, and although it is composed mainly of ready-made modules, it requires time for programming and debugging. And this indicator will allow in a visual mode in the tester to simulate manual trading at different phases of quotes history, and immediately assess whether the strategy is worth attention or not. (If anyone does not know how to trade manually in the tester, to speed up manual testing as compared to demo account: I manage orders in my EA using global variables, I start the EA in the visualization mode, attach the needed indicator to the window, set the suitable speed of running, as soon as the visual combination of signal lines is formed to open a position - I press pause, F3, calling the window of global variables, correct the value of the global variable for the position in the window, close the window, shoot the cursor.
6) To return to the essence of the subject, I experimented but I still do not get stuck in formalizing the dependence of the indicator operation on quotes changes, I think the phases of the market and indicator signal lines are visually similar, but there is an imperceptible shift and the logic of TS works differently. I was expecting a brainstorming session and set this task, but it seems it is not interesting for many people.
7) And in general, it's time to go on holiday, to the sea, in the sun, because I'm tired, tiring myself out, training bulls and bears.
1) I did not bring an example with the filling of the tank by accident. On the one hand I was trying to help you see the problem from a slightly different point of view, on the other hand - the process of filling the tank with liquid is similar to filling the deposit with profit and can relatively easily be modeled in TraceMode. TraceMode is indeed positioned as a system for industrial production management, but I recommended it because it applies exactly the way you see the problem. After the virtual "industrial" process control, you could proceed to real bidding by writing a program for MT. But there is one But, pardon the pun.... More on that later in the text.
2) I've already regretted mentioning mashups in my previous post. Of course, I know that you know that everyone knows about the mashkas. The backbone of my narrative doesn't hang on to them, of course. :)
3) Let's go on about filling the barrel. Although the fluid flow is turbulent, it can still be described with sufficient accuracy for practical purposes by means of formulas, formalised. That is, any number of sensors (temperature, flow rate, pressure, the presence of impurities, the appearance of cavitation, etc.) can be installed to ensure control of the process with any (almost) accuracy. But the trouble is - the fluid flow can be formalised, the quote flow cannot. After all, this is exactly what those who want to get control of the process of making money in the markets by building a TS with feedbacks are trying to do, intentionally or not.... Therein lies some irony in the barrel example.
Regardless of the number of TS parameters, I see only two ways to build adaptive TS. I wrote a few posts above:
a) changing parameters continuously (continuity of course implies changing TS parameters on every bar, less discreteness is impossible to achieve due to the discrete nature of the signal itself)
b) parameters change after a certain period of time.
And without any feedback. This is where you have to dig in a different direction - adaptation by choosing among many parallel alternatives (there are many ways to choose). There is no goal of identifying the cause (finding effective feedbacks).
4) .....
5) I observe a growing number of those who switch to "handheld" and "semi-handheld". Well, this phenomenon is likely to be periodic in nature. Then there will be a shift in the opposite direction. Everyone makes their own choices. Not bad or good, just their own.
6) See the previous points. I don't think the topic is of no interest to anyone. It's just that people for the most part don't know what to latch onto. Look at the number of downloads from codebase - the last on that list would be libraries. People like ready-made solutions. And you have to dig and dig in your subject and it is not clear where to dig.
7) You need a rest. And the menagerie isn't going anywhere. :)
PS I reread what I wrote - a lot of notes. Sorry. But in general, this topic is unlikely to be ever discussed so that it can be said in a nutshell. Good luck!
1)I don't quite understand your statement. Do you mean that technical analysis does not work in principle and its use is useless, and therefore the adaptation of TS to market changes is not possible? I am not talking about the fact that I cannot make profit, but about the fact that I cannot model the system of feedbacks adequately tuning the TS according to the changes.
2)And what special tests are you talking about, can you give us some examples?
1) No, you're talking about the fact that you can't make money, and then blah blah blah and something about "feedbacks" adequately rebuilding the TA for "ongoing changes". Once again, "ongoing changes" is a random wandering - even backwards, even direct, rebuild or not, nothing will work.Those who write that they earn on this pair, for example, and not only on this one - are they simply liars, or they are still lucky, or? are sodruzhniki TC, who maintain the atmosphere of faith and hope for the bright future.
2) I personally like Dickey-Fuller test (DF) or expanded Dickey-Fuller test (ADF), you can do it with autoregressive AR(p) or complete ARIMA(p,d,q) model - with this you can determine the process that we observe: random walk, mean-reverting, explode process; is there a linear determined trend or drift in the random walk.
The first time I found a breakeven in my trading system, it was the first time I tried to do it in a demo account.
I have found a breakeven in my TS since the first time - I've tried it on both demo and micro accounts - most of the time I breakeven, or let's say, if I got a profit, I only got it under manual control.
I have chosen the forex for the Expert Advisor with a breakthrough of the channel in the low TF in the major trend of the high TF - the only problem is the flat, at the moment I am trying to fix the losses in the flat, but I think it would be better to try working with pending orders
the trend on a higher TF is determined by the indicator on the basis of the wagons, the only problem is correct order closing - I open and close one order after the signal to enter the opposite one, but I suspect a completely different indicator is needed for the exit or maybe I should work with a higher timeframe