Machine learning in trading: theory, models, practice and algo-trading - page 788
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
The main thing that would not silence then and did not run away with stuffed bags from the forum :)
And where is the topicstarter? )) Maybe he already ran away ... A long time ago.
Alexei is right. The 5th bar prediction may give +100 pts and the 10th one - 0 pts. As a result we miss the trade. Or the forecast on the 10th bar may show you +100 pts. But on the 5th bar it may be -100 pts and you will catch a stop. It is necessary to forecast all of the bars from 1 to 10.
The disadvantage is increasing complexity of the network, because there are more exits and as a result the time of calculations is longer. Also it is necessary to increase number of neurons in hidden layers...
Read topkstarter's blog - there too regression and a lot of smart ideas.
Well piss, I thought that here are professionals or at least experienced people gathered.... I'm writing from the phone, but when I get to the computer, you will dance with me...... on the first bar, I predict a 10 on the second nine on the third eighth. I see the whole window and tell me how I can get the elk if I see the whole window? well, just curious to know....
Well piss, I thought that here are professionals or at least experienced people gathered.... I'm writing from the phone, but when I get to the computer, you will dance with me...... on the first bar, I predict 10 on the second nine on the third eighth. I see the whole window and tell me how I can get a los if I see the whole window? just curious to know....
maybe they meant to say that by predicting the price at the close of the 10th bar you miss the point that before the close of the 10th bar the price could shoot higher or lower, capturing so to speak some stops.
Well piss, I thought that here are professionals or at least experienced people gathered.... I'm writing from the phone, but when I get to the computer, you will dance with me...... on the first bar, I predict 10 on the second nine on the third eighth. I see the whole window and tell me how I can get the elk if I see the whole window? well, just curious to know....
Yes, it's true.
But there was...
Here's a picture of the situation, just because you saw this drop three bars before, does not mean that you take it into account now.
Well piss, I thought that here are professionals or at least experienced people gathered.... I'm writing from the phone, but when I get to the computer, you will dance with me...... on the first bar, I predict 10 on the second nine on the third eighth. I see the whole window and tell me how I can get a moose if I see the whole window? well, just curious to know....
No, I'm not a professional at all, I'm just here to take a closer look at what's what, maybe I'm asking stupid questions, but it's a way to fathom the truth.
I just think that time doesn't matter here, what matters is what happened during that time. We need to make conditional entries into position, and if we see that the price rose and fell ten times more than the delta on the tenth bar, then this will significantly affect the trading system - why should we wait for 10 bars, if we could take a take profit 5 times more, or close on a stop/no loss, if within 10 bars we dived into a landing with a lot of profit on 10 bars. Maybe it is difficult to implement, then report it, then it will be clear. In the meantime, your comments do not make me understand why we only look at the closing price.
They probably meant to say that by predicting the price of the cloze of the 10th bar you miss the point that before the cloze of the 10th bar the price could shoot higher or lower, hitting some stops, so to speak.
All right.
I believe that independent
1. yes
2. may create misunderstandings, because the output labels will constantly change during training and their fitting. But it is possible to analyze the dynamics of label changes and find the sets with the least amount of fitting
Then Doc is right. The best model of any process is the process itself. Where you suggest stopping somewhere is regularization. But it has never solved the problem of overfitting, just a trick, sometimes a good one. The very idea of a changing teacher seems to me to be flawed for financial series. In genetics, pattern recognition (letters, sounds) is understandable, but in financial rows it should not work.
But I will hope, that I am wrong and you are right.
Well piss, I thought that here are professionals or at least experienced people gathered.... I'm writing from the phone, but when I get to the computer, you will dance with me...... I predict 10 on the first bar, nine on the second, eight on the third. I see the whole window and tell me how can I get a moose if I see the whole window? well, just curious to know....
There is also an objection here - since the forecast 10 bars ago, there may have been some events that fundamentally change the price movement. For example 10 bars ago you predicted growth, on the 5th bar the news turned the price down and the new forecast must be down - so we can increase the accuracy of the forecast due to taking into account the latest data.
But if the forecast on the 10th bar will always be high enough, then your method makes sense without re-predicting the intermediate window bars. This has to be tried and compared.
Well piss, I thought that here are professionals or at least experienced people gathered.... I'm writing from the phone, but when I get to the computer, you will dance with me...... on the first bar, I predict 10 on the second nine on the third eighth. I see the whole window and tell me how I can get a los if I see the whole window? well, just curious to know....
If you`re predicting the closes, and professionals know, that they get lots by hi-jacks and looses...