The arrival of a new generation of trading software. What should the EA interface be like? - page 3

 

Again, it is worth estimating how much time it takes to write the interface itself, and how much we are going to save on that.

And try to estimate production costs.

That is, assess whether the game is worth it.

 

- in terms of programming: OO (object orientation) will move and elements of FP (functional programming) and ES (expert systems) will be actively used

- visual" application builders will rule

- the era of trade terminals will most likely be over - the current terminal will be a "router" for data and orders to classical applications. The terminal cannot "grasp the immensity".

so I think...

 
Nikolay Demko:

The interface is secondary to the functionality.

Each EA has its own functional needs (some have more than enough standard instances).

There cannot be a universal interface.

The most universal functionality is the basic MQL.

All the rest are the languages of a higher level with respect to MQL.

If MQL is not so complex, why bother to think about building a universal interface language? How much demand for it will arise?

Suppose we have a certain market research. For this, we write an Expert Advisor that is most convenient to change. And it requires some kind of management interface.

The interface itself is implemented according to the needs of management, how much information should be displayed, what commands should be received and upon what events.

Once these questions have been answered, it is easier to create an interface, without any universal tricks.

Otherwise, without any specific binding it will turn out like this:

1) The interface is secondary to the functionality when human intervention is completely ruled out and contraindicated. If human intervention is necessary, the interface is a kind of adapter through which additional functionality in the form of human actions and decisions is connected. The question of whether a robot should be completely independent is debatable, but it is indisputable that at least sometimes a human should intervene in its activities, and this requires an interface.

2. We will not create a special language to create the interface. Developers in the future will create it in a very simple way, without writing program code. This technology will be finished soon. So there is no need to worry about that. Later on, I will describe in more detail how exactly the interface creation will be done. Very easy.

3. Unfortunately, it is very difficult to create a full-fledged interface using conventional methods. Though now there are graphic libraries which considerably facilitate this work...

 
Maxim Kuznetsov:

- in terms of programming: OO (object orientation) will move and elements of FP (functional programming) and ES (expert systems) will be actively used

- visual" application builders will rule

- the era of trade terminals will most likely be over - the current terminal will be a "router" for data and orders to classical applications. The terminal cannot "grasp the immensity".

so I think...

I agree with your point of view. A visual application builder can include not only work with graphics, but also work with the logical core of the program. The technology is essentially very simple. All that is needed is to index (number) function calls and put these indexes in a three-dimensional kernel (array). Some of the functions will check events, others will perform procedures and calculations. This kernel must be "attached" to a logic engine that will move along the "perimeter" of events, selecting the required logical chain.

The compilation of such a kernel can also be done in a visual environment.

 

Strange...

What's it all for?

It seems to me that the author himself doesn't understand what he wants.

I think the author himself does not understand what he wants.

My Expert Advisors, for instance, don't have any interface or settings. Once you run it, it works. Just watch when the TS stops working - and remove the "offenders" by putting new ones in....

 
George Merts:

Strange...

What's it all for?

It seems to me that the author himself doesn't understand what he wants.

In reality, everyone would like to have an Expert Advisor with only one button - "chop dough". And all this "interface functionality" is just a tricky one.

Let's say my Expert Advisors have no interface or settings at all. You start them up and they work. Just watch when the TS stops working - and remove the "wrongdoers" by putting in new ones....

Let's answer the question: is the "chop dough" button a 100% guarantee of its "chopping"? Obviously, if we create one "chop dough" button instead of an interface and keep pressing it, we will not make more money. A sensible trader understands that underneath the button there must be code to support the main function.

In reality, such a button requires an ideal functionality, so perfect that the AI itself cannot compete with it, because the AI is a copy of a human mind, not God, and therefore its effectiveness will not be higher than that of an ordinary trader, and for such a button you need a system of error-free forecasting. Until we have it, it is naive to expect an effective system running on one button, and constantly "chopping dough" without any additional settings, without a user interface, without human intervention. To disregard the user interface as an additional tool for program management in real trading, based on faith in such a magic button, is simply not wise.

 
Maxim Kuznetsov:

- in terms of programming: OO (object orientation) will move and elements of FP (functional programming) and ES (expert systems) will be actively used

- visual" application builders will rule

- the era of trade terminals will most likely be over - the current terminal will be a "router" for data and orders to classical applications. The terminal cannot "grasp the immensity".

so I think...

Yes, probably there will appear a 3S (ServerSideScript) - development of OCO order ideas and moving elements of HFT directly to servers. The technology allows this, the only problem is the protocols and the language. That is, the execution logic will become really distributed
 
Maxim Kuznetsov:
Yes, there will probably be some kind of 3S (ServerSideScript) - development of OCO order ideas and moving HFT elements directly to servers. The technology allows this, the only problem is the protocols and the language. That is, the execution logic will become truly distributed

I'm sure the big players' robots have a complete interface, customisation system, visualisation and other things. The robots are of a higher level. I would like to reach them.

Everything human beings do is subject to change and improvement. I think further market trading will evolve from HFT machines to smart robots.

Trading with a "smart" robot, applying a complex strategy, adjusting its performance, collecting and analysing statistics, is much more interesting than setting up a machine, running it and then just hoping to make a profit.

 
Реter Konow:

I'm sure the big players' robots have a complete interface, customisation system, visualisation and so on. The robots are of a higher level. I would like to reach them.

Everything human beings do is subject to change and improvement. I think further market trading will evolve from HFT machines to smart robots.

Trading with a "smart" robot, applying a complex strategy, adjusting its performance, collecting and analysing statistics, is much more interesting than setting up a machine, running it and then just hoping to make a profit.


The big players trade with the simplest, trend-following robot.

They want iMA = 20 and that's it. Then they average out and close at a loss of 20%.

 

In abstract, about the "robots of the big players" - they don't have robots :-) at least in our EA sense...there are policies (a kind of mega-strategy), there are analytics centres, there are activators (traders + classic robots) implementing some goals.