Machine learning in trading: theory, models, practice and algo-trading - page 1871

 
Maxim Dmitrievsky:
Here is the question: average the signals on all bars, or check the conditions for each. The second one is more labor-intensive, but the quality of signals can be improved.

It is more difficult to average to a certain limit, and check the conditions if they are out of limits. I would use width and angle. The maximum angle of the trend returns quickly, risks)))).

And it's not a fact that counting for every bar will improve the quality of the signal... Then the average conditions in relation to / for each tick.

 
Mihail Marchukajtes:
Max, if the prediction is within the next two hour bars, this is the most important thing. Think about what their faces will be when they will be in a negative MO :-) WAHAHAHAHAHAHAHAHAHAHA. I'd like to see the look on their faces. Well, what do these organizers wanted, when they have modern technology. this is one. Otherwise, you can work with simple positions, but in cases up to a change in the sign of the predicted bar. That may be quite frequent in the conditions of these TS, but the price of an open position and flip will be important here. Here you need to understand if the market gives you a good price, you need to go in, but if it is not giving you a good forecast, then it's better to skip this forecast and orientate on the next.... In general it's like this, if I understand correctly the last periods of your discussions. IMHO of course!!!!

Limits and FxSaber))))

 
Valeriy Yastremskiy:

Limits and FxSaber))))

They won't always work, which will also be a skip. So everything is the same...
 
mytarmailS:

Mud, do something already, at least on one cluster, at least one rule, enough of the lyrics already, here no one will advise you anything in your project. Make something that works, and then, if it fires, you can think about improving the

I have no doubts that it will work.

how did you cut it by hours?

i sometimes have something missing or the time is shifted somewhere

i.e. some clocks are kind of out of sequence, for certain intervals. I have already checked all the code, there are no errors
 
Maxim Dmitrievsky:

And where did you get the quotes from, not through the mt5 package?

And how did you thin out the hours?

i sometimes miss something or the time is shifted somewhere

I have already looked through all the errors in the code. I already checked all the code, there are no errors

If it is not in the expected place, then wait so long, if not, then skip.

apparently the time range does not coincide with the real one and a half two ranges or more)

 
Mihail Marchukajtes:
They will not always work, which will also be a skip. So it's the same thing...

skipping a profitable option is not wrong.

 
Valeriy Yastremskiy:

mixes must be corrected by logic, if not in the expected place, then wait so much, if not, then skip.

apparently the time range does not coincide with the real one and a half two ranges or more)

If you want to check it line by line, you need to look for the discrepancy. At the transition clock sometimes large gaps, clearly the hours are not one day shifted

May be because of holidays or sessions opened at different times (holidays or something else).

 
Maxim Dmitrievsky:

-and where did you get the quotes from, not through the mt5 package?

-- and how did you thin out the hours

-- I wrote a script on mt4, it simply writes quotes in tcht when the candle closes, I read it with the "P" and do what I want

- I wrote a function, which converts the time to hours, and then it's clear what to do



_________________

Amazing video, and the man himself too


 
mytarmailS:

-- I wrote a script on mt4, it just writes quotes in txt when the candle closes, I read it with "P" and do what I want

-- I wrote a function, which converts the time to hours, and it's clear what's next



_________________

Awesome video, and so is the man himself.


try 5 or 6 clocks over 2-3 years, will there be breaks in the curves? sharp transitions

like this crap.


 
Maxim Dmitrievsky:

Try 5-6 watches over 2-3 years, will there be breaks in the curves? abrupt transitions

like this kind of crap.

I do not know, I erased that script the same day)

There shouldn't be such a logic
Reason: