Machine learning in trading: theory, models, practice and algo-trading - page 244
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
Heh heh. how misguided you are.
https://nplus1.ru/news/2016/11/03/glasses
This is quite an exotic feature of deep neural networks, journos love to trash such fleas, like the joke about neural networks mixing up negroes and gorillas, but it is not statistically significant, the important thing is that only with MO you can get something from the market, more randomness, besides specifically for bundles of time series with so much noise as in the market, neural networks are not the best choice.
Not bullshit, it's the only thing we have. TA is also MO, just a very inefficient, profaned form of it.
PS: the weather can be predicted quite accurately (70-80%) for the next few days, the face of a person in makeup can also be recognized no worse than a person does.
Bullshit when applied incorrectly, I've explained in a post why bullshit. And I have nothing against MO tools themselves, they are the future in terms of AI and information processing.
Have you seen the TV show "Spot on"? Appearance modification is not limited to square glasses alone. The market behaves in such a way, you can say, that it is not recognized and it succeeds 99% of the time.
It's bullshit when applied incorrectly, I explained in my post why it's bullshit. And I have nothing against MO tools themselves, they are the future in terms of AI and information processing.
Have you seen the TV show "Spot on"? Changing appearance is not limited to square glasses alone.
However, most of the time the jury will accurately recognize actors in makeup.
I wonder how human pattern recognition is superior to machine pattern recognition. Is the human brain a more advanced neural network? More traceable parameters? A wider experience base?
Is it possible to bring a modern computer to this level, or the hardware limitations will not allow?
If we consider the show "Toe-to-Toe" then:
1. The man is not familiar with the speaker participants can not recognize anyone, but certainly all recognize make-up people and distinguish them from nagrimirovannymi.
So, two types of experience are needed to accurately recognize makeuped people:
(a) The experience of perceiving human faces without makeup.
b) The experience of perceiving human faces in makeup.
2. To recognize the speakers, one needs a third type of experience: knowledge of the faces and other physiological parameters of the participants. One needs not only the knowledge of appearance, but also the movements, the voice tones, the mannerisms, the actions...
I can conclude: If a machine is trained to recognize all of these parameters, it too can sit on a jury).
It's bullshit when applied incorrectly, I explained in my post why it's bullshit. And I have nothing against MO tools themselves, they are the future in terms of AI and information processing.
Have you seen the TV show "Spot on"? Appearance modification is not limited to square glasses alone. The market behaves, you could say, in such a way that it is not recognized and it succeeds 99% of the time.
In that case you're right, but I wouldn't formulate it "right/wrong", the question is rather about mastery of the application of MO, in the subtleties, for quite complex data with a lot of noise. Take for example the same numer.ai, most have >0.69 including me, it's 54-55% of the image, but there are those who have <0.6 it means about 70% of the image, not sure that they act "correctly", I mean using standard means.
However, most of the time, the jury accurately recognizes actors in makeup.
I wonder how human pattern recognition is superior to machine pattern recognition. Is the human brain a more advanced neural network? More traceable parameters? A broader experience base?
Can a modern computer be brought up to this level, or will hardware limitations prevent it?
In most cases, participants don't have the goal of being unrecognized at all, the goal is to be like the chosen image. But this is just an example of how a person's appearance can change, if you want you can change your appearance completely so that even your mother would not recognize it, up to and including surgery on the vocal chords changing fingerprints and replacing irises with donor eyes. The marketplace is why it changes, so as not to be recognized, not so as to be recognized in a different image.
But what remains constant about people with altered appearances? What is invariable is that they remain human, which means they have two legs, two arms, etc., that is, the attributes of a human being remain and do not change. Nor does a person's general behavior change; they won't eat out of a cat's bowl, for example.
That is, instead of detailing the traits, we should generalize the traits, identify the unchanging traits, and exploit them.
In most cases, participants do not have the goal of being unidentified at all, the goal is to look like the chosen image. But this is just an example of how a person's appearance can change; if you want, you can change your appearance completely so that even your mother would not recognize you, up to and including surgery on the vocal chords to change fingerprints and replacing the irises with donor eyes. The marketplace is why it changes, so as not to be recognized, not so as to be recognized in a different image.
But what remains constant about people with altered appearances? What is invariable is that they remain human, which means they have two legs, two arms, etc., that is, the attributes of a human being remain and do not change. Nor does a person's general behavior change; they won't eat out of a cat's bowl, for example.
That is, instead of detailing the traits, we should generalize the traits, identify the unchanging traits, and exploit them.
You're right again, in algorithmic trading with MO it is more important to have data and attributes than classification itself, all this jerk with candlestick patterns is nothing more than noise.
In most cases, participants do not have the goal of being unidentified at all, the goal is to look like the chosen image. But this is just an example of how a person's appearance can change; if you want, you can change your appearance completely so that even your mother would not recognize it, up to and including surgery on the vocal chords to change fingerprints and replacing the irises with donor eyes. The marketplace is why it changes, so as not to be recognized, not so as to be recognized in a different image.
But what remains constant about people with altered appearances? What is invariable is that they remain human, which means they have two legs, two arms, etc., that is, the attributes of a human being remain and do not change. Nor does a person's general behavior change; they won't eat out of a cat's bowl, for example.
That is, instead of detailing the traits, we should generalize the traits, identify the unchanging traits, and exploit them.
I think we are going in the right direction. We just need to be clear about the difference between human perception and machine perception.
Human perception is perfected throughout life. Human beings enrich their experience with everything they come into contact with. Thinking helps him to build logical structures and abstract images. Human brain neuronet has an enormous potential of learning and development. 3.
2. The machine is inherently dependent on the creator.
3. its experience is invested within it and limited to one specific area.
4. The machine is constrained by its hardware limitations. This also limits its experience.
I think to train modern neuronet is like to train an insect - a lot of work, but not much use. However, maybe if we change the approach or make more advanced computers, it will be better.I think we are going in the right direction. We just need to be clear about the difference between human perception and machine perception.
1. Human perception is perfected throughout life. Human beings enrich their experience with everything they come into contact with. Thinking helps him to build logical structures and abstract images. Human brain neuronet has an enormous potential of learning and development. 3.
2. The machine is inherently dependent on the creator.
3. its experience is invested within it and limited to one specific area.
4. The machine is constrained by its hardware limitations. This also limits its experience.
I think training a modern neuronet is like training an insect - a lot of work, but not much use. However, perhaps if you change the approach or make more advanced computers, it will be better.
I say we should simplify, generalize, and you say we should complicate, detail. Why should we try to match human perception? How much use is human perception in the market, if manual traders were losing 20 years ago by reading the ticker, and they are still losing now using technical analysis and MODs.
Let's think logically:
To simplify something, we need to thoroughly understand the complexity of this thing. To know its structure. I see the process as a complication, detailing and simplification. And so every cycle of development. Raising to a new and new level.
Machine learning is a tool in the hands of an algotrader, and this tool must be improved in any case.
As for the effectiveness of neural network market forecasting - this is debatable. I think with the right approach, efficiency can be obtained.