Machine learning in trading: theory, models, practice and algo-trading - page 2142

[mytarmailS]

Evgeniy Chumakov:

Run this ZZ in the NS.

What do you want me to predict? Normalized time or normalized price?

[mytarmailS]

Evgeniy Chumakov:

Preferably the price, time as an additional row (whether it affects the forecast or not)

In general NS will reveal something interesting from these two series or not.

the importance of the signs

forecasts after the vertical line

blue forecast

[mytarmailS]

Evgeniy Chumakov:

i.e., the time series has no effect on the prediction?

And predict the time series

blue forecast-is that normal forecasting or not?

Didn't mess up the direction of the row this time?

predicts AFIGUALLY!!!

but let's turn on our heads, if we give a number of peaks and troughs, do we need time to understand that after the peak will be a trough, and do we need AMO to understand the same?

[mytarmailS]

time

[mytarmailS]

Evgeniy Chumakov:

Can you make a file like the one I posted, but add a forecast column to it?

And how did you do the normalization?

Here's all the code, from reading the file, to training the model, to writing

dt <- read.table(file = "C:\\......\\EURUSD_zz_step_100.txt",sep = ",",header = F)

x1 <- dt$V4
x2 <- dt$V5  #  time

X <- cbind(  hank(x1,10) , hank(x2,10) )
Y <- c(diff(x1),0)

tr <- 100:5000
ts <- 5001:length(x1)

rf <- randomForest(Y[tr]~., X[tr,] ,ntree=100)
pr <- predict(rf,X[ts,])

DT <- tail(dt,length(PR1))
dt2 <- cbind(DT,PR1,PR2)
dt2<- dt2[,-6]

write.csv2(dt2,file = "C:\\.....\\Desktop\\ddd.csv")

Is it so complicated ...

[mytarmailS]

Evgeniy Chumakov:

Well yes, it was even easier here https://www.mql5.com/ru/forum/86386/page2094#comment_19133494

got it )))) accepted )

[mytarmailS]

Evgeniy Chumakov:

Something in your file row does not coincide with the forecast (values + -)

Show me a picture

[Vladimir Perervenko]

mytarmailS:

Vladimir, do you know what sickness ZZ from TTR package

Sometimes he draws such inadequacies.

And in general the more I look at it, the more it seems to me inadequate

It is a normal, adequate ZZ. But to draw ZZ on EURUSD with a 9 pips knee? I have not had such artifacts for a long time.

пример
zz <- TTR::ZigZag(HL = cbind(d$X.HIGH.,d$X.LOW.) ,change = 0.0009,percent = F) 

1. Check the class HL. It has to be either matrix or xts. What are those dots at the end of the name? d$X.HIGH.

2. I haven't used a knee less than 25 pt(4 digits) on this symbol in a long time. Play with the size of the knee.

[mytarmailS]

Vladimir Perervenko:

Normal, adequate ZZ. But to build a ZZ on EURUSD with a knee of 9 pips? I haven't had such artifacts for a long time.

1. Check the class HL. It has to be either matrix or xts. What are those dots at the end of the name? d$X.HIGH.

2. I haven't used a knee smaller than 25 pt(4 digits) on this symbol in a long time. Play with the size of the knee.

Ok, I'll figure it out.

And this is just the way R-ka calls my variables when I read from tcht

Evgeniy Chumakov:

Blue row, red forecast.

Well, the forecast is known that point before.

Although it should have recorded taking into account, eh...

[Valeriy Yastremskiy]

Vladimir Perervenko:

Better than 144.

Possible. The logic is approximately a couple of hours, a day, a week, a month, a quarter, a year, 10 years.