Machine learning in trading: theory, models, practice and algo-trading - page 2545
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
Not denying the waves... Waves are secondary, primary are impulses of news / events. Which turn into waves. But it is a level problem, what will be the waves from a pebble flying along such and such a trajectory, weighing such and such a configuration. Waves depend on all three and not only three parameters of the pebble, but also on external influences not only on the pebble...
They do.
And here is how machine learning can be applied to the news, rumors and statements of politicians.)
I've been reading the MoD nonsense here for several years. It's a funny thread.
Sorry for the off-topic. Went away quietly.
That's the way it is.
And here's how machine learning can be attached to the news, rumors, and statements of politicians.)
Apparently only as of late. When somewhere has popped the NS should learn to join the movement.
Apparently only as of late. When it goes somewhere the NS would have to learn to join the movement.
Then the whole disadvantage of MO is that the NS should either refuse to learn or be very late like most technical indicators. Then what is the point of learning if it won't fit the new realities of the market?
I am not against the NS. I use NS myself, but without training. I use strict ratios. NS is used as an element in TS for confirmation of other signal sources. Can-do not-wait.
Output result is in percent. This is actually the power of directional movement. It is also convenient to observe the wave reversal or, as some people say, the trend. This approach is available for any TF. Training and retraining of the NS is not required, since price behavior is identical in any TF.
Then the whole disadvantage of MO is that the NS must either refuse to learn or be very delayed like most technical indicators. Then what's the point of learning if it won't fit the new market realities?
I am not against the NS. I use NS myself, but without training. I use strict ratios. NS is used as an element in TS for confirmation of other signal sources. Can-do not-wait.
Output result is in percent. This is actually the power of directional movement. It is also convenient to observe the wave reversal or, as some people say, the trend. This approach is available for any TF. Training and retraining of the NS is not required, since price behavior is identical in any TF.
This is how easy the result maylook.
Finding similar situations in the history, the MO will report that, for example, in 7 cases out of 10 similar ones there was a growth. There is a lag, but not a big one. The MO remembers what happened and does not wait for the MA to turn. In general, patterns like head and shoulders the MO will find and suggest the most probable result. The problem is that the new data turn out to be about 50/50.
And we won't have time to follow the news even when sitting in front of the terminal. Usually the news appears when the movement has already started a few seconds ago.
I like the definition of "neural network based database" well or forest/bust.
Finding similar situations in the history, the MO will report that, for example, in 7 cases out of 10 similar ones there was a growth. There is a lag, but not a big one. The MO remembers what happened and does not wait for the MA to turn. In general, patterns like head and shoulders the MO will find and suggest the most probable result. The problem is that the new data turn out to be about 50/50.
Well, we won't have time to follow the news even when sitting in front of the terminal. Usually the news appears when the movement has already started a few seconds ago.
I have a prognosticating duty on the NS in general.
You can't expect the NS with the MO to react on an instantaneous price change, because the market price is not a perfect sinusoid. Trained on one data will not be suitable for new ones, moreover in a short time interval for the signal.
By the way, lately the head-shoulders are not the same as they were 15 years ago). The sharks have learned how to skillfully catch liquidity on them. Life flows, everything changes. The peeps are getting craftier. New ways and tricks for making a profit are needed. Today it's short-term trading. I work in this direction.
Here are a couple of quotes from the article:
"The peculiarity of artificial intelligence is that the technology is not capable of navigating new non-standard situations. If an abnormal situation occurs in the market, the model is unlikely to suggest the best way out. The pandemic is a prime example of this. The Organization for Economic Cooperation and Development (OECD)cites data that, according to aBank of England survey, during this period about 35% of banks experienced negative consequences from the operation of the AI model based on machine learning. This is primarily due to the fact that the pandemic caused changes in many macroeconomic indicators, which became the parameters involved in the development of the models.
Given these characteristics of artificial intelligence, many financial institutions do not give it complete freedom of action. At Sberbank, for example, AI is not allowed to control trading robots directly. Rather, it acts as a "smart" assistant and gives the trader recommendations on how to configure the execution algorithm. At the same time, the final decision is always made by a human."
--------------
"Thanks to algorithms, trades can be conducted automatically, with minimal trader involvement.In 2018, about 80% of trades in the U.S. stock market were almost entirely controlled by machines,Jupiter Asset Managementtold us. "
-------------
NS isn't AI yet, but it's already an element of it, with even bigger problems in trading.
Here are a couple of quotes from the article:
"The peculiarity of artificial intelligence is that the technology is not capable of navigating new non-standard situations. If an abnormal situation occurs in the market, the model is unlikely to suggest the best way out. The pandemic is a prime example of this. The Organization for Economic Cooperation and Development (OECD)cites data that, according to aBank of England survey, during this period about 35% of banks experienced negative consequences from the operation of the AI model based on machine learning. This is primarily due to the fact that the pandemic caused changes in many macroeconomic indicators, which became the parameters involved in the development of the models.
Given these characteristics of artificial intelligence, many financial institutions do not give it complete freedom of action. At Sberbank, for example, AI is not allowed to control trading robots directly. Rather, it acts as a "smart" assistant and gives the trader recommendations on how to configure the execution algorithm. At the same time, the final decision is always made by a human."
--------------
"Thanks to algorithms, trades can be conducted automatically, with minimal trader involvement.In 2018, about 80% of trades in the U.S. stock market were almost entirely controlled by machines,Jupiter Asset Managementtold us. "
-------------
NS is not yet AI, but already an element of it, with even bigger problems in trading.
Who downloaded what interesting, is it worth further refinement?
Maybe there is something in the comparison of the forecast of some main component with the real component in the future.
If we remember the forecast, we compare the current component with the one that "should have been" according to the forecast for correlation ...
If the correlation is strong then we kind of trade...
Just an idea of where we can move...
It's like this, as long as the correlation between the forecast and reality is strong then we try to believe the forecast...
Forecast by the second main component from the moving average, the first one deleted Te detrendil
Maybe there is something in comparing the forecast of some major component with the real component in the future.
Those remembering the forecast we compare the current component with the one that "should have been according to the forecast" for correlation ...
If the correlation is strong then we kind of trade...
Just an idea of where we can move...
It's like this, as long as the correlation between the forecast and reality is strong then we try to believe the forecast
I made a forecast by the second main component from the moving average, the first one deleted Te detrendil
The question is how many times the forecast will work.
https://www.mql5.com/ru/forum/330111/page2#comment_16078211
https://www.mql5.com/ru/articles/318 (5 estimate of forecast accuracy)