Neural network - page 13
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
It is useful to think about such a question. Suppose we have a neural network which has been trained on certain input signals to give correct answers 90% of the time (soros rests). Obviously, no, because the information about what inputs are required and how to interpret outputs is not stored in the network, but in the mind of its creator. Thus, although the network is trained, it turns out to be useless. Once again. An NS is just a tool (imho, not the best and not the worst of others available), to have it and know how to use it are completely different things.
Brrrr. So what good is it if it gets 90% of the answers right? So you took a probabilistic grid that does not contain contradictions.
Find the right columns (choose the right attributes - the task at a time) - and it gave 90% due to the absence of contradiction in the data.
What's next? The example is incorrect. Even a 3-step self-consistency-verification-sampling-out-of-sample check does not get it right.
Brrrr. So what good is it if it gets 90% of the answers right? So you took a probabilistic grid, which does not contain contradictions.
We found the right columns (choosing the right attributes is a one-time task) and it gave 90% due to the absence of contradiction in the data.
What's next? The example is incorrect. Even a 3-step self-consistency-check-sample-out-of-sample check does not yield what is needed.
I'm just explaining why pre-processing is necessary and which direction it should go and which it should not.
Ps to offtope: you do not understand even I understand!!! (с)
for alsu :
Tell me - have you managed to make a self-learning grid yourself, cram everything essential into it, and profit from it?
I do not understand the volumes. Do you get the volume data from the Volume indicator?
Can't find any inputs in there...
It's simple. What you see are the inputs...
Hello to all! I've been working on Trading Solution for a long time - it's not bad, but rather dumb (sorry!). Its bars start not from right to left but from left to right. The thing is, I've developed an input system that's killer and unique. But now I cannot use it on the history because I cannot attach the neuronet in MT4 and I cannot reproduce the system of input signal in soluchene. Can anyone help me with this?
Before transmitting (or after receiving), change the direction of indexing in the array using the
By the way, pay attention that a regular array and a timeseries array have different indexing directions,
see Array operations section for help.
Good afternoon. This may be off-topic, but I would still like to ask experts. I want to know if it is possible with the help of some software, for example Staistica (maybe even with its neuronet) to determine the regularity of trades. There is a bot that trades and has the results of 3 months of trades (entry time, buy price, profit), we have an approximate idea of what the system is based on, but the program may understand and display the result that trades have the following conditions Are there such programs?
Good afternoon. This may be off-topic, but I would still like to ask experts. I want to know if it is possible with the help of some software, for example Staistica (maybe even with its neuronet) to determine the regularity of trades. There is a bot that trades and has the results of 3 months of trades (entry time, buy price, profit), we have an approximate idea of what the system is based on, but the program may understand and display the result that trades have the following conditions Are there such programs?
There is no much difference between the software and the program. It's just that proprietary software has support and is more GUI. The free one, in fact, the same, but in a kind of kit, following the principle "build it yourself if you understand it".
As for regularities, any neural network, even the most parfunky one, can find them if they are really present. Another question: are the found features of objects clear regularities or only special cases?
Good afternoon. This may be off-topic, but I would still like to ask experts. I want to know if it is possible with the help of some software, for example Staistica (maybe even with its neuronet) to determine the regularity of trades. There is a bot that trades and has the results of 3 months of trades (entry time, buy price, profit), we have an approximate idea of what the system is based on, but the program may understand and display the result that trades have the following conditions Are there such programs?
The grid will find the existing skew in the data, there is no doubt about it. The only question is how long the skew will last. You are not the only one who will pounce on it. And very quickly the market will become symmetrical again. Take a look around the site at the perceptron EA charts. At the beginning they go up and then down. Of course you can try to make an adaptive perceptron, but that's another story.There is not much difference from the software. Simply, proprietary software has support and is more GUI. And free software is essentially the same, only in the form of some kind of "build it yourself, if you understand it" kit.
As for patterns, any neural network, even the most parfunky one, can find them, if they are real. Another question: are found features of objects clear regularities or just special cases?
But it is very complex, I know that we can build corners with certain MAHs and calculate something else from those corners and then they are compared between two currency pairs, and there are some other signals there too, so, may the software understand what corners are, how they are built and what is calculated from those corners? And if it's real, then how to do it in Statistica 6.1 (or other software), in what form these data should be loaded there. For example I loaded the history of two pairs of currencies, and transactions as a load (that it is a variable?) Or just a set of data to download and neuro understand that this is a time series that here is the price of buying this currency, and those are the other.....and so on, if someone can suggest how to implement please help. >>Thank you.