Machine learning in trading: theory, models, practice and algo-trading - page 1529
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
With examples on the sine wave
hmm.... am i being followed? i want a grid but with a non-linear step, i want to find the step through a polynomial, look for the polynomial formula in the optimizer ( the polynomial coefficients in the settings)
))))
hmm.... Am I being followed? I want a grid but with a non-linear step, I want to find the step through a polynomial, look for the polynomial formula in the optimizer (the polynomial coefficients in the settings)
))))
By the way, the fuzzy logic is easy to use, the easiest and most effective way in my opinion... in terms of speed and interpretability of results :) You can make a grid or whatever the hell you want
I liked these articles on how to approximate curves with fuzzy logichttps://towardsdatascience.com/a-short-tutorial-on-fuzzy-time-series-part-iii-69445dff83fb
hmm.... Am I being followed? I want a grid but with a non-linear step, I want to find the step through a polynomial, look for the polynomial formula in the optimizer (polynomial coefficients in the settings)
))))
no need for a polynomial
it is enough to reduce the step if the profit is negative
if (pips < 0) {
step *= Math.Exp(pips / Kstep);
}
Is it reasonable to start to develop an algorithm based on a neural network that will input normalized indicator data and markers that will signal the presence of some patterns?
Another idea is to make a system of multiple neural networks that will be divided into categories (different types of indicators, patterns, etc.) and the outputs of these neural networks pass through the final neural network to make a decision?
Looking back at the thread of this forum, which stretches back to 2016, I would like to ask about the success of the "Random Forest" algorithm.
It may seem like lamer questions (it probably is) and they have already been raised, but I would still like some answers)
Thanks in advance)
Hello) I would like to ask a couple of questions related to algotrading. I am a Python/Go developer myself, my knowledge of trading is at the level of having read a couple of books on technical and candlestick analysis.
Is it reasonable to start to develop an algorithm based on a neural network that will input normalized indicator data and markers that will signal the presence of some patterns?
Another idea is to make a system of many neural networks that will be divided into categories (different types of indicators, patterns, etc.), and the outputs of these neural networks should be passed through the final neural network to make a decision?
Looking back at the branch of this forum that stretches back to 2016, I would like to ask about the success of the "Random Forest" algorithm.
It may seem like lamer questions (it probably is) and they have already been raised, but I would still like some answers)
Thanks in advance)
Hello). Welcome to the club of grail seekers)).
The main problem is "fitting" the model to the historical data. The rest is easier here, like money management, model selection, programming, etc. Statistics will come in handy for you.
And about neural networks, scaffolding is quite possible, the question is the right application...
Hello) I would like to ask some questions about algotrading. I am a Python/Go developer myself, my knowledge of trading is at the level of having read a couple of books on technical and candlestick analysis.
Is it reasonable to start to develop an algorithm based on a neural network that will input normalized indicator data and markers that will signal the presence of some patterns?
Another idea is to make a system of many neural networks that will be divided into categories (different types of indicators, patterns, etc.), and the outputs of these neural networks should be passed through the final neural network to make a decision?
Looking back at the branch of this forum that stretches back to 2016, I would like to ask about the success of the "Random Forest" algorithm.
It may seem like lamer questions (it probably is) and they have already been raised, but I would still like some answers)
Thanks in advance)
Forest just indicators will not take out, if you do not know the pattern. You can try boosting with crossvalidation and early stopping, something newer and cooler. Betting on neural networks also makes little sense - you won't find so many indicators, they will correlate with each other. Emphasis on the oversampling of instruments (currency pairs or whatever). Crypto also makes sense, it is not as effective as forex pairs. For crypto, you can play with arbitrage and with the cup.
Is there a point to start to develop an algorithm based on a neural network in which normalized indicator data and markers that will signal the presence of some patterns will come in?
It depends on what you mean by that. If you are a coder-researcher by vocation, in fact, then it is a very interesting task, which then will not be ashamed to boast about, if you follow through, but as an attempt to "earn on the Internet", alas, will not work, at least not as soon as you think, maybe if you are a genius then in 10 000 hours, maybe in 20 000 ... but most likely never(? but most likely never((
Another idea is to make a system of many neural networks that will be divided into categories (different types of indicators, patterns, etc.), and let the outputs of these neural networks pass through the final neural network to make a decision?
You have invented stacking, but the problem is not in the algorithms, but in the data, you need more and better data for profitable algorithmic trading.
Looking back at the branch of this forum that stretches back to 2016, I would like to ask about the success of the "Random Forest" algorithm.
Random forest is one of the most effective machine learning algorithms for most tasks, for individual cases you can get fractions of % additionally with boosting or their combinations "boobag, bagbu", but again, it's all about data, and it is expensive and collecting it in the right quantity/quality for algotrading is also a separate task.
Hello) I would like to ask a couple of questions related to algotrading. I am a Python/Go developer myself, my knowledge of trading is on the level of having read a couple of books on technical and candlestick analysis.
Is it reasonable to start to develop an algorithm based on a neural network that will input normalized indicator data and markers that will signal the presence of some patterns?
Another idea is to make a system of many neural networks that will be divided into categories (different types of indicators, patterns, etc.), and the outputs of these neural networks should be passed through the final neural network to make a decision?
Looking back at the branch of this forum that stretches back to 2016, I would like to ask about the success of the "Random Forest" algorithm.
It may seem like lamer questions (it probably is) and they have already been raised, but I would still like some answers)
Thanks in advance)
I'll tell you this, I'm quite satisfied with trading networks, and I've been searching for 15 years, but when I found them I spend no more than 2 hours a week. I mean optimization of Expert Advisor and everything related to it. If you are a Python developer, then like any other developer, I advise you to thoroughly and thoroughly learn the domain in which you are going to work. The market is not only quotes the market is mostly about PEOPLE!!!!! What a statement, your thought should be like a shot, otherwise you won`t have time to drink :-)
And so far the question is still unresolved as to what is the reason for my success.