Machine learning in trading: theory, models, practice and algo-trading - page 2002

 
mytarmailS:

Yeprst))))) who does that ))))


Everyone does as he pleases. I originally wrote where to count from.

 
Evgeniy Chumakov:

массив[0],[1],[2],[3],...[n]

you need to predict cell 0.

ForecastStart is not a range, but an offset. That is, I start with cell 1. Where x[ForecastStart + 9] = 10 array cell.

So the range is from cell 1 to 10.

it was predicting well because it was predicting past )))

)) no one does it that way ))

Max was doing the wrong model too...

 
Evgeniy Chumakov:


Everyone does as he pleases. I originally wrote from where to count.

What is your data? How did you obtain it? From the prices?
You didn't use a zigzag, by any chance?
 
mytarmailS:


the new values should be at the end of the file, not at the beginning)) no one does that))


It's more convenient for me to work in an array, and everyone does as he or she wants.

 
Evgeniy Chumakov:

It's more convenient for me to work in an array, and everyone does it the way they want.

That's why everything turns out wrong, because everyone does it the way he wants, not the way it's supposed to be done.

I'm sure Max made the same mistake and didn't read your data inside out, but did it normally according to the standard...


here are the real cuts

          Reference
Prediction  -1   1
        -1 371 209
        1  236 373
                                          
               Accuracy : 0.6257          
                 95% CI : (0.5975, 0.6533)
    No Information Rate : 0.5105          
    P-Value [Acc > NIR] : 8.14 e-16   

real cuts and no cuts)) here's reality knocking.

 
Here are two files, the first as I posted. The second is the right one.
Files:
 
mytarmailS:

That's why everything turns out wrong, because everyone does it the way he wants, not the way it's supposed to be done.


It's nobody's fault if people are too lazy to read.

reality comes knocking - well, it's not my fault))

 
Evgeniy Chumakov:
Here are two files, the first as I posted. The second as it should be.

I have already done everything, the result is above

 
Evgeny Chumakov:

It's nobody's fault if people are too lazy to read.

Reality knocked - well, it's not my fault)))

)))

 
mytarmailS:

)))


The question is different.

Here you are (if you can call it that, correct me if what) did not know that file upside down, succeeded in predicting the next values, but the program does not know where the past and where the future. It only sees the numbers. Or is it a mystery?

We look at the graph and we don't know reliably what will happen on the right in the future. And if we look at the same graph backwards, we know the past (which is on the left side of the graph) because we can look, but if we hide part of the graph on the left, can we reliably tell what should be there?