Machine learning in trading: theory, models, practice and algo-trading - page 439
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 don't get it, is Yuri gone?
It depends on your belief in the afterlife.
just sharing my experience.....
1. It is possible to look at the market analysis from a different angle, I'm no longer trying to find patterns that will show me where the price will go or levels from which the price will bounce, I could not find anything really worthwhile,
But I can look for levels where the price will go for sure within several days, for example, I can look for levels at which the price will definitely go in the future, maybe today, maybe after tomorrow, but I will definitely go there... I found such levels, they work with 100% probability - those are always...
2. The market is not a time series, it's an event model... ... Therefore, common methods do not work, starting from stupid oscillators/indicators and ending with networks, etc... The information is not processed correctly from the beginning, the market is not BP.
p.s. of course this is my own story, but maybe someone will find it useful
just sharing my experience.....
2. The market is not a time series
))) No one, no one, no one is convinced of anything...
Don't argue, you are both my students, or rather DEAD students, both of you were not up to the granite of science, both betrayed their teacher, now you are talking nonsense and strive for ruin, for your own destiny
Nevertheless, the fact remains that machine learning methods, consciously used, only proved the high degree of market efficiency and the lack of meaningful trading opportunities for the "average" (normal) player using public data and popular MO methods.
This will especially hit suckers and scammers, although of course convinced losers and professional near-market players serving them by facts alone will not change their minds, but nevertheless with time Pinocchio and Alice and Basilio, will be much less,
Nevertheless, the fact remains that machine learning methods, consciously used, only proved the high degree of market efficiency and the lack of meaningful trading opportunities for the "average" (normal) player using public data and popular MO methods.
This will especially hit suckers and razzers, though of course convinced loosers and professional traders serving them only facts cannot change their minds, but nevertheless with time Pinocchio and Alice and Basilio, will be much less,
Nevertheless, the fact remains that machine learning methods, consciously used, only proved a high measure of market efficiency and the lack of meaningful trading opportunities in it for the "average" (normal) player using public data and popular MO methods.
I don't know what you mean by effective, but let's just say :
1) the market is driven by an exchange robot that reacts to the balance of those willing to buy and sell an asset.
2) if there are more buyers, the market will go down to fulfill them, because there are more buyers, and therefore the exchange will take more commission from them than from the sellers.
3) To predict the movement on the market, you need to predict the actions of the robot,
In order to predict the actions of the robot, you need to predict the actions of market participants (traders, robots (the majority)). This robot will be their counteragent.
4) to predict the actions of the majority, you need to understand what they all have in common, and the only thing they have in common is history, everyone uses the history one way or another, starting from the ordinary manual trader and ending with algotraders who use history to race their robots
5) So if at the moment we have such a pattern, that was repeated on the history and it was "profitable" to trade long on the history, most likely the robots will buy the moment, then they will pour "buy" liquidity into the market, thereby making the market fall
6) So, the MO more or less helps to find such patterns, to which the crowd will react
I do not know, unfortunately or fortunately, but all written in this thread are not proof of anything. It's a mystery where you found any facts that prove anything...)
Lazy to look who exactly, but at least 2-3 people above wrote that they were disappointed in MO, not much came out attractive, went back to patterns and oldskulu. And I understand them, as well as myself have not received from the MO, or rather received proof that the market is efficient At least with the data and methods of MO available to the masses.
I do not know what you mean by efficient, let's just say :
1) the market is driven by a robot that reacts to the balance of those who want to buy and sell an asset.
2) If there are more buyers, the market will go down to fulfill them, because there are more buyers, and therefore the exchange will take more commission from them than from the sellers.
3) To predict the movement on the market, you need to predict the actions of the robot,
In order to predict the actions of the robot, you need to predict the actions of market participants (traders, robots (the majority)). This robot will be their counteragent.
4) to predict the actions of the majority, you need to understand what they all have in common, and the only thing they have in common is history, everyone uses the history one way or another, starting from the ordinary manual trader and ending with algotraders who use history to race their robots
5) So if at the moment we have such a pattern, that was repeated on the history and it was "profitable" to trade long on the history, then most likely the robots will buy the moment they will pour "buy" liquidity into the market, thereby making the market fall
6) Thus, MO more or less helps to average find such patterns, to which the crowd will react
The assumption is beautiful, but not empirically correct, otherwise it would be enough a simple strategy signal reversal in the MO to get the Grail, but it doesn't happen.
PS: "efficiency" is unpredictability when price takes all the information into account.
PPS: But unlike old-school (TA, patterns, etc.) with an overfit on the tester, properly prepared MO, gives an honest result, "cuts the truth of the womb", but not everyone has the guts to accept this result.