"New Neural" is an Open Source neural network engine project for the MetaTrader 5 platform. - page 65
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
Tex. Personally, I know almost nothing about networks. From what I know, I think they are of no use to the market.
I want to make sure that this is true or to make sure that it is not true at all.
At the stage of acquaintance and mastering, I do not want to touch market quotes. I want to try the network on a sine wave with something I have defined; I want to try it on an hourly temperature chart, gradually incorporating data on pressure, time of year, humidity, wind strength and time of day.
With 95% accuracy I will indicate the temperature of the next hour based on these data, knowing their correlation. Will the grid be able to forecast one hour ahead, or two hours ahead, or more?
What you're giving me is nothing for the mt 4. You know how this box works, I don't.
You won't be able to check if the grids are suitable for the market, if you research day-ahead temperature forecasting, even if you succeed and become cooler than hydrometeorological center.
Although, your position is clear - you want maximum simplicity and at the same time versatility in the applicability to a variety of data: this task-objective is the first and foremost of the project.
You can't make sure if the grids are right for the market if you're researching the possibility of predicting temperatures a day in advance,
In other words, if it cannot extrapolate even for one hour , say, having all data, it does not mean, that it is not suitable for market analysis and I do not know what I will feed it with?
Maybe yes, maybe no.
But if you want to make sure of the suitability/unsuitability of nets for the market, torture the market, not the weather. :)
I'm kidding of course. Simply, if you have spent a considerable amount of time on making your complex predict the weather, you will be surprised when it turns out that what you have been doing for years - doesn't work in the market. But that's not the network's fault. For the market it was just necessary to build a different model. (By model I mean the final configuration of the settings and parameters of the neurocomplex).
Maybe yes, maybe no.
But if you want to make sure of the suitability/unsuitability of nets for the market, torture the market, not the weather. :)
I'm kidding of course. It's just, having spent some time on weather forecasting, you will be surprised to find out that what you were adjusting for years is not working in the market. But that's not the network's fault. For the market it was just necessary to build a different model. (by model I mean the final configuration of settings and parameters of the neurocomplex).
No no no . There are two options here. The first - to sink completely into the construction of the box, for this you need to have knowledge of your (those who here can not compromise, find the common denominator and finally begin) level.
The second is not to dive in, but to accept the box "as is". The second is not to dive in, but to accept the box "as is" by reading the manual for a few buttons and adjustments.
That is, if it cannot extrapolate even for one hour, for example, this chart having all data, it does not follow that it is not suitable for the market analysis, moreover not knowing what I will feed it?
Misha the question of weather prediction is not so simple, scientists develop a model, they enter the necessary parameters into the model (temperature, humidity, pressure, ...).
After that a map is built for the entire earth, data on these parameters are recorded for each point, and the model builds a forecast for each specific point.
But you want to give the model data from one point of the map and receive the forecast, it is impossible because the accuracy of this forecast will be influenced by other points of the map.
The most you get is an approximation of diurnal fluctuations, since the sun rises and sets with an enviable constancy.
But a decrease in temperature because of a thunderstorm front will not work, because the data that you feed does not contain information about its approach, it is contained in other points on the map (which you have hidden from the neuronics).
The second is not to take the plunge, but to accept the box "as is". by studying the instructions for a few buttons and adjustments.
Ideally, this is how it should work.
True, in addition to the button "make me feel good" will be another button to be pressed first - "learn to make me feel good".
Such a complex can be loaded into the firmware of a kettle, an iron, a TV, an iPhone, a car navigator, and even a nitendo. Vibrators and massagers will also be able to be loaded into them. And, after training, each of these devices will behave in a way that will give the owner the greatest pleasure.
Now I'm almost not kidding.
Ideally, this is how it should work.
But in addition to the "make me feel good" button, there will be another button to be pressed first - "learn to make me feel good.
Such a complex can be loaded into the firmware of a kettle, an iron, a TV, an iPhone, a car navigator, and even a nitendo. Vibrators and massagers will also be able to be loaded into them. And, after training, each of these devices will behave in a way that will give the owner the greatest pleasure.
I'm almost not kidding now.
So maybe it's easier to implant electrodes into an individual's pleasure center?
Imho, it's easier than explaining to every dummy that he shouldn't throw the iron at passersby during training.
I can already see the lawsuits against the iron company for the fact that their improperly trained hot iron stroked the owner on the back :)
Misha the question of weather prediction is not so simple, scientists develop a model, they enter the necessary parameters into the model (temperature, humidity, pressure, ...).
After that a map is built for the entire earth, data on these parameters are recorded for each point, and the model builds a forecast for each specific point, studying changes over time.
But you want to give the model data from one point of the map and receive the forecast, it is impossible because the accuracy of this forecast will be influenced by other points of the map.
The most you'll get is an approximation of the daily fluctuations, since the sun rises and sets with an enviable consistency.
But a decrease in temperature because of a thunderstorm front will not work, because in the data that you supply there is no information about its approach, it is contained in other points of the map (which you have hidden from the neuronics).
What you are talking about is a long-term forecast.
Once again, there's a 95% chance I'll give you the next hour's forecast myself.
By the way in 98 I communicated with the employee of the Petersburg hydrometeocenter and they made forecasts for 2-4 weeks ahead looking in the archive of the previous years with the maximum coincidence for the current day.
But it does not matter.
Weather one hour ahead is very simple. Weather one day ahead with a 55% probability is easy. That is why I personally want to familiarize myself with the grid on this data.
So maybe it is easier to implant electrodes in the pleasure center of the individual?
We have half the country living with electrodes ((
Especially on Friday there is an increased demand for electrodes, contacts and earthing
So maybe it is easier to implant electrodes in the pleasure center of an individual?
I think it's easier than explaining to every moron that he shouldn't throw them at people while learning the iron.
And just think what we are raising our hand to! After all, it will be possible to load our complex into the aircraft's onboard computer as easily as into the massager. We have already developed universal interfaces (or almost have).