Machine learning in trading: theory, models, practice and algo-trading - page 889
![MQL5 - Language of trade strategies built-in the MetaTrader 5 client terminal](https://c.mql5.com/i/registerlandings/logo-2.png)
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
During training it got weights and offsets for neurons, and according to them it calculates output on new data.
So she got the weights just by statistics. Otherwise it can not distribute weights without knowing what was on the last bar. Well, or the networks are more primitive than I thought...
So she got the weights just by statistics. Otherwise it can not distribute weights without knowing what was on the last bar. Or the network is more primitive than I thought...
Certainly by statistics, as are the forests, as are your arrays.
Yes, I mean, then it makes no sense in what chronology to submit information for analysis.
Yes, I mean that then there is no point in which chronology to submit the information for analysis.
Almost none, only because of the lock minima or recurrence.
but they also break out of the lock minima by themselves normally, if anything you can retrain a few times and compareHow is the best tree in a random forest selected or an average tree created?
Am I correct in assuming that the fewer rules the tree has, the more stable the model?
Then what ratio is considered normal? I have 403933 lines for training, formed 69779 rules, so it turns out that for every 5.79 lines 1 rule, which seems to me a lot, or is it normal? If you look at the reliability, the ratio changes there - in the higher side, which means that the distribution is not uniform, but how to see what it is there ...
Here's the first test subject ready - training 2015-2016, and from 2017 pure trading on selected tree rules - didn't lose - already good?
Against trading without NS - training (ugh - tuning and optimization) 2016-2017
Still don't understand the best way to do it - I ended up selecting rules and turning them into code - very painstaking manual work... Need some kind of automation of the process.
And this is 2017-2018 by tree rules for input, but with filters not yet integrated into the training dataset
We need some kind of automation of the process.
Got it! The program has grouped by rules lines with different input data (and I said that you can unload the results as a file, but there are some sets of variables to each line, statistics and rule number), and now we need to find which variables in these lines are the same - this will be the rule!
I'll sleep, and think how to organize it better - any ideas - write!
Got it! The program has grouped by rules lines with different input data (and I said that you can unload the results as a file, but there are some sets of variables to each line, statistics and rule number), and now we need to find which variables in these lines are the same - this will be the rule!
I'll sleep, and think how to organize it better - any ideas - write!
Try to apply the optimizer on the forest from my article, it may help you improve the results!
I have a code if you need it.