Time of writing the advisor - page 6

 
Maxim Romanov:

You can program anything, but you have to develop an algorithm first, and that's a big part of the job. In essence, you have to turn what you see first into formulas and logic, and then you have to program it. And the first is very often underestimated.

A programme is not a wizard. Even such "simple" things as levels or SL and TP cannot be determined programmatically.

Their fixed values must be defined or set manually. And what will happen when TP is set at 50 points (for 4 signs), and the price reaches 48-49 points and goes in the opposite direction?

How would your robot react? Of course, it would not close the position. The same will happen with levels, with trends, with reversals, etc.

How will you teach your robot what to do in such cases?

 
Petros Shatakhtsyan:

The software is not a wizard. Even "simple" things like levels or SL and TP cannot be determined programmatically.

Their fixed values have to be defined or set manually. And what happens when TP is set at 50 pips (for 4 digits), and the price reaches 48-49 pips and goes in the opposite direction.

How would your robot react? Of course, it would not close the position. The same will happen with levels, with trends, with reversals, etc.

How can you teach your robot what to do in such cases?

There is an algorithm for that, no matter where the price reaches, there will always be a situation where it did not reach 1 point, whether it is a human or a robot. Algorithm can be created for everything, the only question is how complicated it is. Impossible there is no such thing as "haven't figured out how", it can be. People have made self-driving cars, there are more variables there than on the market. Even the driver has been automated, and once upon a time it was impossible.

The problem of building levels needs to start with defining what levels are. They do not know what they are, they talk about levels, but what they are, why they are there, why they should work, nobody says because they do not know. But if you don't know, then what to automate? Of course, you cannot create a program if you don't know what you are automating. Why should SL and TP be in this place? You have to start there, the rest is all expressed in formulas and logic.

 

I'll add to the discussion.


1. Question to the customer: How formalized was the ToR? I've encountered this:
  • I want a robot to break a level.
  • OK, what is the algorithm for determining levels?
  • Well, here are the screenshots, so...
  • I see, so there is no algorithm and I have to come up with it myself? Then $200 just for the algorithm (this includes three or four reworkings in the process of agreeing on it) + the robot itself.
  • REALLY??? IT'S SO EASY!!! 100$ for everything!!!

Not your case?

2. Question for the developer: what is the structure of the programme? The best thing in large projects (and in any) is to divide it into elementary, and, most importantly, independent of each other blocks, OOP is of help here. After that, if the structure is correct, it is a pleasure to make changes. For example, to replace one principle of input/output definition with another, just write a new class or function and change it from the previous one, and you don't have to think about all other 1000+ lines of code. That's why, as I wrote earlier, the first release is about a month, and it's already a finished robot backbone, in which, after that, only individual blocks are added/refined/changed/deleted.

Here too, as an example:

  • The previous developer has gone into the sunset, fix the robot, I need this, this and this, and I need to fix a bug here as well.
I look at the code and there is 2500 lines of procedural code, and everything runs in a row; if functions are defined, they only need to open/close/calculate the number of orders. I ask for the initial requirements and the whole amount of work, up to the first tests, takes about 8 hours and 1000 lines of code.
 
Maxim Romanov:
That's what I'm asking, what's the volume of the terms of reference. It took me about 7 months for a robot with a ToR of just over 40 sheets. But you have to understand that freelancers don't just do your work.

Wow, where are the kind customers who pay for this amount of work?

 
The situation is very interesting. Here you are writing for an order, something went wrong, the customer never agrees to pay even $10 extra (or with great difficulty they are pulled out, even when the situation is obvious). And here you are - a man paying 800 p. per hour, and the programmer works 10 hours a day.
 

Then you will find out what the courses are for 2 t.c.s., about which it was strictly forbidden to tell, and that will be the end of it. That's the whole point of the topic.

 
Dmitry Fedoseev:

Wow, where do such kind customers live that pay for this amount of work?

So I am a customer)
 
Petros Shatakhtsyan:

The software is not a wizard. Even "simple" things like levels or SL and TP cannot be determined programmatically.

Their fixed values have to be defined or set manually. And what happens when TP is set at 50 pips (for 4 digits), and the price reaches 48-49 pips and goes in the opposite direction.

How would your robot react? Of course, it would not close the position. The same will happen with levels, with trends, with reversals, etc.

How do you teach your robot what to do in such cases?

My scalper has SL and TP only in case of nuclear war, the robot closes there somehow, so I don't ask it. And in general, complex TP is nonsense. I've seen a lot of them in my life.

I won't forget one hot Estonian guy, who wanted to display about 2000 settings, out of which 1950 were absolutely useless. When I told him that it's not technically possible, I got one answer - you're a great programmer, come up with an idea.

I said to him - Matty, you have a naked martin in your strategy, you will inevitably lose your $40,000 soon! You have to think not about the beauty on the screen, but about how not to get lost. Better buy yourself a tractor in a farm and forget about forex. He was a farmer then, no higher education and generally maths at multiplication table level.

- You don't understand, the more I lose the more I win later, that's such an Estonian wisdom! We retreat first, then we always win!

I don't think there's any need to state the bottom line. It all turned out to be the Russians' fault, that is me personally. Forty tons down the drain.

----------------

Keep it simple. I'm working on some tricks and I'm slowly pulling the scalper up from Monday till yesterday's Thursday, 4 days. It's a bummer, but if I can get some stability I'll get something. Real.

1


But guys, I often wonder why we are fidgeting in this forex, do we really...

2

 
will do, for a moment, and then there will be a bitter hangover...
 
A freelance service should have a section on writing the TOR from the customer's words.
If you explain clearly in words what is needed, all that is left is to apply it to the "geography of the graphic".
I've always been pretty good at reasoning logic. Perhaps, with enough understanding of strategy, I could apply all this to graphics.