Thank you Maxim for sharing such a masterpiece!!! Took me an entire day of work an review of some maths (and it worth it) to fully understand the power of this tool. A new door has been opened. And for the question above : add "void" at the beginning of the line
It works exactly until 2019.06.01. Author published it June 2019. After that date until 2022.08.01 it generates only losses.
Thank you for contributing your skills to the reader.
Based on this and his other articles, the programmer is certainly one of the best who publishes at Metaquotes. Unfortunately, this does not improve the performance of his EA.
Hi, this is some kind of research different econometric methods, to show what an entropy does and how it save information in time series, plus optimization with logistic regression. But to use it in raeal trading need more research I think and more different tests.
I understand that in the indicator normalisation of diff. prices is carried out on the basis of once calculated mean/std. But how correct is it to train the model by a previously unknown mean of the whole hist_display range? Using a sliding window for normalisation is not more plausible?
babbetta89 #: I understand that in the indicator normalisation of diff. prices is carried out on the basis of once calculated mean/std. But how correct is it to train the model by a previously unknown mean of the whole hist_display range? Using a sliding window for normalisation is not more plausible?
Need to remember :) yes, you could try your second option.
add type
And for the question above : add "void" at the beginning of the line
< deleted - no insult >
It works exactly until 2019.06.01. Author published it June 2019. After that date until 2022.08.01 it generates only losses.
Thank you for contributing your skills to the reader.
Based on this and his other articles, the programmer is certainly one of the best who publishes at Metaquotes. Unfortunately, this does not improve the performance of his EA.I understand that in the indicator normalisation of diff. prices is carried out on the basis of once calculated mean/std. But how correct is it to train the model by a previously unknown mean of the whole hist_display range? Using a sliding window for normalisation is not more plausible?