NS + indicators. Experiment. - page 3
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
Hello dear sirs!
I apologise at once for the fact that this is supposedly my first post. In fact, I have been "here" for a long time. Only recently I changed my computer and with it the operating system. Therefore, all my logins and passwords have gone into oblivion.
Now to the point. I looked at this article too. The article is like an article. I looked at it from the point of view of my own problems of working with NS. And, in fact, I found something for myself. Of course, the authors overdid it a little from the point of view of principle of the problem. But, gentlemen, let us use our own brains always and not only from 8:00 to 17:00. What am I talking about? Well, the fact that the article is an article (and no more than that), that the author describes his own vision of the problem and ways to solve the problem. And here we go... I understand you are human, but try not to hastily criticize any "work" you may encounter on your way. Kudos to you for expanding your point of view with your second post on the subject of this article. Humanly speaking, it's perfectly understandable, but SK and KimIV's "behavior" is not (understandable though, of course =). SK and KimIV have their own sites that they "author", but "having" their own sites nowadays is not an indicator of anything, especially of their credibility. OK SK, at least he read the article. A KimIV didn't even read it, he just "spat" at the authors.
For a first introduction to the concept of "neural networks", the article may do the trick (although a methodologically correct textbook explains the concept better). Therefore, in my opinion, Prival is right when he says: "It's advertising in a pretty package that gives the false sense that everything is so simple and easy there".
It's just an article, not a textbook. Whether it's good or bad is a second question. I wanted to hear something of substance, not a meeting of the literary circle.
As far as I still remember this forum has something to do with trade, and the theme is about neural networks.
There is another point. "Classical", shall we say,neural network theory is somewhat different from the application of neural networks to financial markets, and even more so if it is a forecast and the operation of the forecast in the future. ....
Hello dear sirs!
I apologise at once for the fact that this is supposedly my first post. In fact, I have been "here" for a long time. Only recently I changed my computer and with it the operating system. So all my logins and passwords have gone into oblivion.
That's how you can recover them through your mailbox. You remember your mailbox, don't you?
Hello dear sirs!
I apologise at once for the fact that this is supposedly my first post. In fact, I have been "here" for a long time. Only recently I changed my computer and with it the operating system. Therefore all my logins and passwords have gone into oblivion.
That's how you can recover through the mailbox. You remember your mailbox, don't you?
I tried it three times. A temporary password (from the box) is sent. None of my past logins matched the password. =((
Tried it three times. A temporary password (by box) is sent. And none of my past logins to this password did not fit. =((
So look up your exact login on the forum....
I would encourage you to think about what you are doing. After all, feeding NS just by trying different indicators is almost the same as trying all known combinations of indicators, and see what happens. Everything is determined by the architecture inside. And there is only one correct input, and that is the quote flow. All the rest is a transformation of this stream.
And you should first try to feed the NS with a stream of quotes, and then write about what is feeding the network. It is always necessary to convert the stream of quotes, otherwise you will get gibberish.
Definitely!) No chance to put quotes in the NS in a bare form, they are noisy, or the network should have a mechanism that has nothing to do with it.
I tried too once to use readings of similar indicators and approximately in the same form for entries:
1) understand well what this or that indicator shows, how it counts and what it takes into account;
2) their (inductors') readings for inputs are also sometimes better to convert/normalize, and very often these indicators appear to be unnecessary "spacers" between the price and the real input.
At the end of the day these two statements led to the situation when smoothing(digital filters and MA) and 33 are the only indicators we use, and reasonably their product is online after serious processing.
By the way, I have a friend who is an elliotic type and I've been telling him all the time that waves are his personal subjective view of the market and they in no way give stat advantage in trading. But if ZZ will give more or less happy results, I will have to bow to him when I meet him.
Mm-hmm. Let's give it a try. By the way, my normalisation is automatic. Please don't think I'm giving raw distances in pips for ns input... I just forgot to specify. My apologies. (corrected in the topic).
By the way, I have a friend who is an elliotic type and I've been telling him all the time that waves are his personal subjective view of the market and they in no way give stat advantage in trading. But if ZZ will give more or less happy results, I will have to bow to him when I meet him.
I recently experimented with ZZ in Day Trader's Neuroshelle. I fed the normalized difference between the price and several fixed exrema ZZ to input PNN (classifier). And also tried ratios of differences (i.e. harmonic models if you want). NS finds truths on a limited time interval. I can't say it's a grail, but the system is in profit on data it hasn't seen.
I recently experimented with ZZ in Neuroshelle Day Trader. I fed to the PNN (classifier) input the normalized difference between the price and several fixed ZZ extremums. I also tried ratios of differences (i.e. harmonic models if you like). NS finds truths on a limited time interval. I won't say it's a grail, but the system is in profit on data it hasn't seen.