Machine learning in trading: theory, models, practice and algo-trading - page 2203
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
This comrade has been gnawing at the crumbs that have fallen from the table of the smart man (Semen Semenych's cluster analysis) for more than a decade. In fact, these clusters are just an attempt to "invent" risk-neutral portfolios for currencies. That is, just another homemade bicycle that won't ride because the markets aren't stationary. So, it quite logically follows the intention of this character to make money by selling fakes for the sake of which the whole forum is covered with all kinds of trash. Administration does not prevent it for obvious reasons.
You`re right, Alexey!
Bad reputation is reputation too.) And in some cases even better than a good one.
Never mind, this nonsense will never end, the moderators are latent. All topics are caked with these characters. If anything, I've got the trolley in my profile, for contact. And you should start to study MO seriously, from A to Z, there are a lot of new topics. Don't get hung up on technology, don't look for anything - you won't find it. No tsos, channels, or other toys for the mentally retarded. Don't waste your time, study MO seriously. It will come in handy, if not for trading, then for an additional profession. Enroll in courses. You think you know something about MO? It's just beginning. The course will help you get your mind right. Digging around on your own is hard. R you know well. They'll give you the math you need on the course. And not the Vorontsovs on YouTube, but courses from real practitioners.
Max, I understand everything, and maybe I will go to some courses, it certainly will not prevent me....
But about the market, we have different points of view, you rely more on algorithms and the newer they are the better, you bet that the MOE will do everything by itself, I'm more interested in digging in search of the essence of the process, some pre-processing data, new types of data representation, it seems to me that if you make the right data, create a proper approach, any AMO copes with the task ...
Each of our approaches has the right to life, and in general it's cool that we do not go the same way, in the future we can exchange experiences and enrich each other .....
You succeeded with VAE, I succeeded with levels (exact entry points on the basis of regression trees), isn't it wonderful?
On the charts extrema are suspiciously accurately defined - are these calculated points on the lines/channel?
These are calculated horizontal levels + entry filter.
I have come up with some "know-how" on how to calculate levels and how to filter them, although everything is so primitive and intuitive that I want to scream "why didn't I understand it before".
But nevertheless, the technology is still raw, we can still improve it, but it already gives good points on the OOS
If you want I can write you what to do, a text file or something like that...
Max, I understand everything, and probably I will go on some courses, it certainly will not prevent me....
But about the market, we have different points of view, you rely more on algorithms and the newer they are the better, you bet that ME will do everything by itself, I am more interested in digging in search of the essence of the process, some pre-processing data, new types of data representation, I think that if you make the right data, create the right approach, any AMO will cope with the task ...
Each of our approaches has the right to life, and in general it's cool that we do not go the same way, in the future we can exchange experiences and enrich each other .....
You've got something with VAE, and I have something with levels (exact entry points on the basis of regression trees), isn't it wonderful?
These are calculated horizontal levels + entry filter.
I have come up with some "know-how" on how to calculate levels and how to filter them, although everything is so primitive and intuitive that I want to scream "How could I have never understood it before?
But nevertheless, the technology is still raw, we can still improve it, but it already gives good points on the OOS
If you want I can write you what to do, a text file or something like that.
Write to everyone. Not many people will get it anyway. But we can talk about it.
Write to everyone. Not many people will get it anyway. But we can talk about it.
OK
But I will not write to all, I will make a file with explanations and send in a private ..., to discuss will be possible here, but without the nuances ...
There is no point in trying to explain things for those who are not familiar with the MO, they will only bring chaos and annoyance to the discussion...
ok
But I will not write to everyone, I will make a file with explanations and send in a private ..., to discuss will be possible here, but without the nuances ...
There is no point in trying to explain myself to those who are not familiar with the MO, they will only contribute to the chaos and irritation in the discussion ...
Vo. And the signal to go all oh... there's no point if you've found how to make a profit.
)
+
These are calculated horizontal levels + input filter
I have come up with some "know-how" on how to calculate levels and how to filter them, although everything is so primitive and intuitive that I want to scream "why didn't I understand it before".
But nevertheless, the technology is still raw, we can still improve it, but it already gives good points on the OOS
If you want I can write you what to do, some text file or something...
Yes, quite interesting, what are these new levels!
Yes, very interesting, what are these new levels!
It's not new levels, it's just normal forecasts, I was doing it about 5 years ago, but now I dug a little deeper...
It's not new levels, it's just the usual predictions, I was doing it about 5 years ago, but now I'm digging a little deeper...
Sometimes it's good to go back to old ideas and solve them with new knowledge!
ok
But I will not write to everyone, I will make a file with explanations and send in a personal ..., to discuss will be possible here, but without the nuances ...
It makes no sense to explain to those who are not familiar with the MO, they will only bring chaos and irritation to the discussion...
The idea of looking for a rebound/reversal is interesting.
1) You make a forecast ahead. How do you trade it? Place orders? As I said earlier, the further forward the forecast is, the closer it is to random.
I would not make the forecast forward, but the opposite at the moment of Y, i.e. at the 0th bar. I would look back, look for a pattern like the one you are looking for and if it was available I would immediately trade Buy|Sell. No pending orders. Get a forecast on the current candle and trade on it.
2) Instead of normalization you can just normalize the price delta from Y (for tree-like patterns you don't need to normalize it). The result will be the same.
3) I think you can't pick individual rules out of the forest. It is necessary to take the average. Otherwise, you will end up with a fitting. We have discussed this with Maxim in the thread on his last article. He did some experiments and confirmed that averaged predictions are worse on test/train, but better on new (exam) data. Read the discussion there.