Full-featured robot for MT5 - page 13
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
Are they going to laugh? It's no big deal. I don't see anything "messy" there either. Everything is relevant, and when a couple of my questions arose along the way were immediately dispelled by further comments - I quite respect that style, and try to write that way myself.
"This is not a trader's idea. A trader's idea is the essence that allows you to make a profit. In essence, a brief description of the technique. Well, let's say,"break through the channel". In your Expert Advisor, I see a bunch of all sorts of techniques. Moreover, they can also switch on and off, and it seems (I haven't found out, but I suspect) according to some quite tricky algorithms.
That's exactly where my doubt lies. I had exactly that, a very sophisticated EA. He was showing great results on a 20 year history. And when I had him for real, he showed rather weak, but still positive results for about a month, and then he lost all his gains in two months. That is when I doubted the usefulness of complex systems. They take much effort to develop, while they work in the same way as the simplest ones.
A "correct" system MUST NOT have parameters. A parametric system, simple or super complex, inevitably REFUSES on history and works well until that history fades into the deep past. :)
Here, history is only needed for TARGETING the algorithms. All "parameters" are taken from the market. One of the basic principles of theanalysis, everything repeats, i.e. patterns or more simply the STATUS is repeated. History is needed to work out ALL situations. If something works wrong in the market - it means either the situation is unfamiliar or freely implemented.
There's always an option (that's what the built-in runner is for) to look for situations that work out and lick them out, rather than counting profits... Profits come by themselves when things are running like clockwork.
Here's a "shoosh", watching the work in the market, found the following "shoosh": the closing system works (and estimated profits on the run) on the close bar. Take and stop are on the distance of volatility from the calculated ones (depending on the level of the flat, volatility and state of the position). So, the market reverses the position earlier due to that it reaches the server stop-loss by a haul or a catch. Hence the divergence occurs during the day's run and the day's run in the market. Corrected... well that's the way it is :)
The "right" system MUST NOT have parameters. A parametric system, simple or super-complex, inevitably REFUSES on history and works well until that history fades into the deep past. :)
Here, history is only needed for TARGETING the algorithms. All "parameters" are taken from the market. One of the basic principles of theanalysis, everything repeats, i.e. patterns or more simply, STATUSS are repeated. History is needed to work out ALL situations. If something does not work correctly in the market - it means that either the situation is unknown, or it is freely implemented.
And how is "debugging algorithms" different from "finding optimal parameters" ? In my opinion, they are the same thing...
Ahhhh... Well if so - then I see the solution to the problem of "having all the current insights into the market" through the constant operation of a bunch of simple systems. We simply observe them - and conclude which systems make the most sense right now. In those directions we look for profit. For example, if we see that trend systems are losing money on some symbol, while flat systems are gaining, then we conclude that it's flat and set variants of flat systems for real trading using this symbol.
That is why I created the League of Trading Systems.
It is interesting to see other variants to solve the same problem ("to have an immediate view of the market"). I'll be watching to see what you come up with.
"Just observe them - and draw a conclusion" that's what a robot does. A human, due to psychological and physiological peculiarities, will NOT be able to make the right decision, and even for a long period of time constantly "observing". A man is a creator, while a robot is an executor. Let the robot "see" if the symbol he sees is in a flat, trend or trend edge and he USES the appropriate subsystem. So our ways are similar :) The shell for your "pile of simple systems", that sees the market and switches systems according to it - the essence of the proposed robot.
How's that for a "trading idea" :) ?
What is the difference between "debugging algorithms" and "finding optimal parameters"? In my opinion, they are one and the same...
Debugging is the answer to the question "what to do in the situation", while fitting is the answer to the question "what numbers to put in here, so that the profit figures are bigger" :)
Well then kind of overcomplicated :)
Only one counter-trend algorithm "somersault" does the following:
1. identifies the trend edge. (long movement relative to the average, increasing counter volume above the average, circular price nearby)
2) Determines the entrance against the trend (sideways movement should start, the signal goes against the trend at the round price, sufficient speed, limits the number of open positions).
3. The AFC closing system is also specific. (Sudden move to profit leads to Takeovers, dragging sideways leads to stop and Takeovers being compressed closer to the open price, smooth move to profit is rejected).
And there are all sorts of little things like CALENDAR - prohibition to trade during certain hours of a dead market, spread control, control of passivity or superactivity of the market, data for N days (here we have 3 days) on averages, maxima and minima... etc... :)
It wouldn't be bad to see the formulas, they often turn out to be quite different from the verbal formulations / wants.
An edge of a trend is an extremum that is identified, after which the price does not change this extremum, and the conditions for an edge of a trend are met). It is the easiest one here. Further it is more complicated. We can look at the time and how far and at what speed the price goes in the other direction. And this is how correctly - what you define it with.
"Just observe them - and draw a conclusion" is what a robot does. Humans, due to psychological and physiological characteristics, will NOT be able to make the right decision, and over a long period of time they are constantly "observing". A man is a creator, while a robot is an executor. Let the robot "see" if the symbol he sees is in a flat, trend or trend edge and he USES the appropriate subsystem. So our ways are similar :) The shell for your "pile of simple systems", that sees the market and switches systems according to it - the essence of the proposed robot.
How's that for a "trading idea" :) ?
Yeah, it's about the same as mine.
But I've never found an algorithm for selecting the TS that I think is better to use at the moment. And this task I solve intuitively.
It will be interesting to see how you do it.
It would not be bad to see the formulas, they often turn out to be quite different from the verbal formulation/wants.
A trend edge, is an extremum that is identified, after which the price does not change that extremum, and the conditions for a trend edge are met). It is the easiest one here. Further it is more complicated. We can look at the time and how far and at what speed the price goes in the other direction. And this is how correctly - with what you define it.
The entire text is in front of you. What formulas?
If the trend edge is defined correctly, the price MUST go against the trend. By definition :) You need to determine the moment to open the position. The ratio of profitable and losing trades is an indicator of correct determination. Some days it reaches 100% with about 10 trades. There are days 0% (it means quantity plus equals quantity minus) but it is rare :) The task is to reduce such days to a minimum.
Yes, it roughly corresponds to mine.
But I haven't found a coherent algorithm for identifying which TCs are best to use at a given moment. And I'm doing it intuitively.
It will be interesting to see how you do it.
Apparently I don't realise what the problem is?
Here we have:
1. an algorithm1 that works well in situation1.
.....
123. an algorithm123 that works well in situation123.
You need an algorithm that detects situations from 1 to 123. Then it's worth nothing to use algorithm33 etc. in situation33. :)
What is done "intuitively"? Is it defining the situation? Or what?
The whole text is in front of you. What formulas?
If the trend edge is defined correctly, price MUST go against the trend. By definition :) It is necessary to determine the moment to open the position. The ratio of profitable and losing trades is an indicator of correct determination. Some days it reaches 100% with about 10 trades. There are days 0% (it means quantity plus equals quantity minus) but it is rare :) The task is to minimize such days.