Machine learning in trading: theory, models, practice and algo-trading - page 250

 
pantural:
Okay, but what if we work as limiters, quoting, standing on both sides like in currency exchanges, selling liquidity? What are the predictions for?
pantural:
That's the answer, we stand limiters on both sides and get a risk-free profit.
All trading requires predictions of future market conditions, no matter trends, pullbacks, MM, arbitrage, volatility, everywhere.
 

it's interesting to look at

https://www.youtube.com/watch?v=p5UngSem9f0

 

Time-consuming.

Can anyone tell me how to bring a matrix, for analysis and subsequent diagnosis through wavelet transforms?

111112468
111123579
1111346810
1112457911
11135681012
11146791113
111578101214
112689111315
1137910121416
11481011131517
12591112141618
136101213151719
147111314161820
158121415171921
169131516182022
271014161719211
381115171820221
49121618192111
510131719202211
61114182021111
71215192122111
8131620221111
914172111111
1015182211111
 

I'm not very good at wavelets, but here, for example, examples with the package dplR were posted, I will rely on them.

1) This is a time series, so the matrix must somehow be converted into one long vector. You have, for example, each matrix row is like a sliding window of length 9, and the window offset on each new row = 0.5. I will add one last element from every other row to the vector. It would be better if you took your vector as it is, without the sliding windows and matrices.
I got a time series like this - 1 1 1 1 1 1 2 4 6 8 10 12 14 16 18 20 22 1 1 1 1 1.

2) What is shaded on the graph is unreliable, and cannot be used to predict the future. The "rightmost on the scale" color in the rightmost unshaded area is green (for periods 2 to 8).
And the answer would be "at the moment, periods from 2 to 8 prevail", which is really useless.)

inputMatrix <- t(matrix(c(1,    1,      1,      1,      1,      2,      4,      6,      8,
                          1,    1,      1,      1,      2,      3,      5,      7,      9,
                          1,    1,      1,      1,      3,      4,      6,      8,      10,
                          1,    1,      1,      2,      4,      5,      7,      9,      11,
                          1,    1,      1,      3,      5,      6,      8,      10,     12,
                          1,    1,      1,      4,      6,      7,      9,      11,     13,
                          1,    1,      1,      5,      7,      8,      10,     12,     14,
                          1,    1,      2,      6,      8,      9,      11,     13,     15,
                          1,    1,      3,      7,      9,      10,     12,     14,     16,
                          1,    1,      4,      8,      10,     11,     13,     15,     17,
                          1,    2,      5,      9,      11,     12,     14,     16,     18,
                          1,    3,      6,      10,     12,     13,     15,     17,     19,
                          1,    4,      7,      11,     13,     14,     16,     18,     20,
                          1,    5,      8,      12,     14,     15,     17,     19,     21,
                          1,    6,      9,      13,     15,     16,     18,     20,     22,
                          2,    7,      10,     14,     16,     17,     19,     21,     1,
                          3,    8,      11,     15,     17,     18,     20,     22,     1,
                          4,    9,      12,     16,     18,     19,     21,     1,      1,
                          5,    10,     13,     17,     19,     20,     22,     1,      1,
                          6,    11,     14,     18,     20,     21,     1,      1,      1,
                          7,    12,     15,     19,     21,     22,     1,      1,      1,
                          8,    13,     16,     20,     22,     1,      1,      1,      1,
                          9,    14,     17,     21,     1,      1,      1,      1,      1,
                          10,   15,     18,     22,     1,      1,      1,      1,      1
                         ), ncol=24, nrow = 9))

#создаём из матрицы временной ряд
for(i in seq(from = 1, to = nrow(inputMatrix), by = 2)){
  if(i==1){
    #первая строка матрицы берётся как есть`
    tsVector <- inputMatrix[i,]
  }else{
    #от каждой строй строки берётся последний элемент
    tsVector <- c(tsVector, inputMatrix[i,ncol(inputMatrix)])
  }
}

#полученный вектор
tsVector

library(dplR)

#рассчёт вейвлетов
waveletObj <- morlet(tsVector)

#показать расчитанные коэффициенты
waveletObj

#график
wavelet.plot(waveletObj)


 
Dr.Trader:

I'm not very good at wavelets, but there were examples with dplR package, I'll rely on them.

1) This is a time series, so the matrix must somehow be converted into one long vector. You have, for example, each matrix row is like a sliding window of length 9, and the window offset on each new row = 0.5. I will add one last element from every other row to the vector. It would be better if you took your vector as it is, without the sliding windows and matrices.
I came out with a time series like this - 1 1 1 1 1 1 2 4 6 8 10 12 14 16 18 20 22 1 1 1 1 1.

2) What is shaded on the graph is unreliable, and cannot be used to predict the future. The "rightmost on the scale" color in the unshaded area is green (for period = 8).


Ooo thank you, I'll look into it!
 

What's the R

Does anyone know if the numbers need to be converted to trigonometric values for the wavelet transform?

Or what do they need to be converted to to start calculations?

What is R?

 
Top2n:

What's R?

Have you tried using google?
 
mytarmailS:
Have you tried using google?
Well, if Radius, then again not very clear. Definitely not a programming language))
 
Has anyone tried to apply knowledge in this area to predicting bets on sports - tennis, soccer, etc.? Or to write bots for poker/pref?
 
Top2n:

What's the R?

The Fourier transform gives results in complex numbers where there arereal and imaginary parts. R in those formulas means to work without imaginary part.