Machine learning in trading: theory, models, practice and algo-trading - page 1105
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
I am interested in a specific algorithm, not general obvious concepts
Alas, already more than has been written for many years on this forum, I doubt that there are any more
Search the forum "broaching a flat", it used to be a hot topic, Kodobase has some examples and there's a regular discussion of flats too... While the market is sideways ))))
And I'm still waiting for someone who will really talk and show developments in MO.
Why wait? People talk to each other is normal. This forum has already discussed everything a hundred times, the search works .... The only thing left is to put it all is not even a bright head, but a firm and assiduous ass!
SZS: And what to expect? Well, perhaps "magic kicks" which periodically distributes@Maxim Dmitrievsky to create an impulse in the right direction with the help of a peculiar manner of communication ))), for what in principle, it is a huge respect! )))
Whatever you call a boat, that's how it sails. (с)
Luge is all about the right number of "s's". (с)
The notion "flat" comes from wave analysis :)
It denotes a "flat" or "sliding flat" correction of the ABC type. It is marked post factum.
It is obvious that in order to mark a flat, it is necessary to draw a wave pattern, because there is a pattern alternation and all patterns are learned by comparison. The notorious"point of reference" should always be present.
I hope that now you have had an epiphany about building castles in the air and fighting windmills.
I hope now you have an epiphany about looking for castles in the air and fighting windmills.
Well, here we go again, Maxim came along and started handing out magic ones. )))
Well, here we go again, Maxim came and started handing out magic ones. )))
As for MO with the rudiments of AI, my opinion now is that it is mostly inductive search, not deductive.
This is when the algorithm itself goes through models that satisfy a certain criterion. For example, stability on new data, regularity. This criterion is external to the model itself, that is, an external censor. The quality of the resulting (found) model depends on the choice of criteria while the input data are indirectly affected :)
When there is no mess in your head, you can take any MO algorithm and twist it, with full understanding, so to speak, of what's going on.As for MO with the rudiments of AI, my opinion now is that it is mostly inductive search, not deductive.
This is when the algorithm itself goes through models that satisfy a certain criterion. For example, stability on new data, regularity. This criterion is external to the model itself, that is, an external censor. The quality of the final (found) model depends on the choice of criteria, and the input data - indirectly :)
Thanks, but I'm still far from it, I still need to read a lot, I'm learning the basics of NS, many new things, which "you can't pick up at first hand".
I think it's some phantasmagoria or utopia at this level of free access materials on AI, and it is unlikely that one or a small group of people are able to develop a fully working or quickly adapting system on their own ... I've written before and I'll write again, if the markets were simple or at least really had an AI model that could profit effectively and for a long time, then Google would be the first to do it - they have a very, very big opportunity.... although who knows, maybe the great google is driving the markets ))))
All I see in the prospect of studying and using AI, is either a rapid reconfiguration of the system (not only retraining and optimization, but also the selection of predictors ) or a semi-automated predictive system.... Such curves with probabilities are drawn by ))))) and according to these curves a decision is made jointly with the user... or vice versa against the user's opinion... I usually don't guess ))))
All I see in the prospect of studying and using AI is either a fast system reconfiguration (not only retraining and optimization, but also selection of predictors) or a semi-automatic prediction system.... Such curves with probabilities are drawn by ))))) and according to these curves a decision is made jointly with the user... or vice versa against the user's opinion... i don't usually guess ))))
Well, what's the phantasmagoria? it's realistic to retrain once per minute bar if the model is fast
google is so capitalized that they do not care about the markets, it seems to me, they are the market
They used to have their own market, based on search queries. Now it's been shut down or something.
Yeah, they shut it down. Google finance is a sad sight now.
https://www.marketbeat.com/press-room/google-finance-changes-and-alternatives/
The notion "flat" comes from wave analysis :)
It denotes a "flat" or "sliding flat" correction of the ABC type. It is marked post factum.
It is obvious that in order to mark a flat, it is necessary to draw a wave pattern, because there is a pattern alternation and all patterns are learned by comparison. The notorious "point of reference" should always be present.
I hope that now you have had an epiphany about building castles in the air and fighting windmills.
And what is correction? It is also a subjective notion, because if it weren't like that there would be some rules that would help to distinguish a correction from a trend. We call those small ones flat, and those big ones trend.
But then again the question is what are small and large movements, in relation to what they are small or large?
In fact these questions give the answer why parametric systems and mo's among them will never work on unprocessed market data
================
Ok, that's all lyrics, now to the main point, why do I need a flat. I have a network that is looking for levels of pd, sp.
Strong levels are formed when the price is in a so-called flat.
eurodollar, 5m, now
these are levels
When the price is in a flat, these are the levels I want to find
And what is a correction? It is also a subjective notion, because if it were not so there would be rules to distinguish correction from the trend, and so correction often turns into a trend and vice versa, and this is because there is no correction, these are completely subjective concepts, there is just movement, movements of different lengths. We call those of small length a flat, and those of large length a trend.
The correction is not an impulse. All these notions are from wave analysis. This notion cannot be objective or subjective, it is objective if it is defined.
Then call it by other words - a corridor or a range. To make it clearer to everyone
Alas, already more than has been written for many years on this forum, I doubt that there is anywhere else
Search the forum "broaching the flat", it used to be a hot topic, the kodobase has some examples, and the discussion of the flat is regular... while the market is flat )))
Why wait? It's normal that people communicate. The search has already been discussed a hundred times on this forum... The only thing left is not even a head but a hard and diligent ass!
SZS: And what to expect? well, perhaps "magic kicks" which periodically distributes@Maxim Dmitrievsky to create an impulse in the right direction by means of the original manner of dialogue ))), for what in principle, to it, the huge respect! )))
I'm waiting, that someone could achieve a result with a lot of clever words expressed in this thread, personally I don't need neither sources, nor algorithms, but the result of MOI work in the form of freezing signal or screenshots for a couple of days. And so far you're just talking and not the point.
About flat and MO, in fact AI in the process of learning will find the necessary probabilistic behavior at the current moment will be a flat or a momentum. So I don't see any sense in writing separate algorithms for determining the flat, it's useless.