Machine learning in trading: theory, models, practice and algo-trading - page 525
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
In fact, the teacher can be anyone, as long as he looks into the future. It is enough even for 1 bar. And the entrances will be exactly those that I wrote about earlier. Believe me, they should be used, but how???? is a different story and that's where the secret of the market lies....
But really - how? Share your secret)
Really.... You just don't know what you're talking about, that's why you're so negative. NS knows how to generalize data and that is enough to have any new example that was not involved in the training. Interpret it correctly....
We went through mathematical modeling back in the 70's.
We came up with the terms NS or machine learning. Well, do this miracle, no one is stopping you.
The answer to all at once about the use of NS.
All predictors are nothing more than some transformations of a time series and do not contain any new information, which is not in the time series, by definition. So, feed the time series itself to the NS and add 1-2 more layers to the NS, and the NS (or another MO system) itself will determine what predictors it needs or does not need, and it will form them itself.
Looking into the future - what is it for? NS can simply answer - to enter the deal or not. In this case, the tasks of the NS are greatly simplified.
Learning with the teacher. Uh-huh, first we have to create an ideal teacher, for which we need to initially know the strategy. So why have an NS (MO) if you already know everything and know how to do everything?
The NS, in my own experience, learns fine with a teacher who doesn't even really know what he wants or what he's teaching. Yes, a slightly different teaching methodology, yes a few more eras, but no more than that.
For this kind of training, you have to imagine the strategy in no more than general terms. The NS will refine it on her own, in the course of the training.
Earlier, in this thread, I already presented the results and graphs obtainedby these approaches on independent samples. So the effectiveness of these approaches has been well demonstrated before.
The answer to all at once about the use of NS.
All predictors are nothing more than some transformations of a time series and do not contain any new information, which is not in the time series, by definition. So, feed the time series itself to the NS and add 1-2 more layers to the NS, and the NS (or another MO system) itself will determine what predictors it needs or does not need, and it will form them itself.
Looking into the future - what is it for? NS can simply answer - to enter the deal or not. In this case, the tasks of the NS are greatly simplified.
Learning with the teacher. Uh-huh, first we must create an ideal teacher, for which we must initially know the strategy. So why have an NS (MO) if you already know everything and know how to do everything?
The NS, in my own experience, learns fine with a teacher who doesn't even really know what he wants or what he's teaching. Yes, a slightly different teaching methodology, yes a few more eras, but no more than that.
For this kind of training, you have to imagine the strategy in no more than general terms. The NS will refine it on his own, in the process of training.
Earlier, in this thread, I already presented the results and graphs obtainedby these approaches on independent samples. So, the effectiveness of these approaches has been well demonstrated before.
And you can show at least one profitable strategy based on IR?
And you can show at least one really working profitable strategy on the basis of MO? So many are able to talk.
I've already shown you on the model. Here is one of the graphs
X-number of deals, Y - total profit. Duration - 3 months.
Real working already in debugging on real time. I do not plan to show it to you.)
Although, here is a debugging deal.
The penultimate digit - 11 - is the profit of the trade. The last one -27 is the maximal profit of the trade.I've already shown it on the model. Here is one of the graphs
X-number of deals, Y - total profit. Duration - 3 months.
Really working already in debugging. I do not plan to show it.)
I have a lot of them. In the tester, each of them has a strictly reversed top-down chart.
I have already shown it on the model. Here is one of the graphs
X-number of deals, Y - total profit. Duration - 3 months.
The real working one is already in debugging. I do not plan to show it to you.)
Although, here is a debugging deal.
The penultimate digit - 11 - is the profit in the trade. The last -27 is the maximum profit in the trade.And you can watch all this in the MT5 tester, in real ticks mode. I can show it without MO.
And you can skip all this on the MT5 tester, in real ticks mode. I can show this without the MO.
I have a lot of them. And each of them in the tester gets a chart that looks strictly the other way round: up - down.
I don't use the MT tester.) And debugging on real-time.
I don't use MT tester) And debugging on real time.
Testing on real ticks, that's what real time is. And very bad that you do not use with the tester.
Testing on real ticks, that's what real time is. And it's too bad that you don't use a tester.