Neural networks in the foreign exchange market

Neural networks - a method of analysis comprising many different units for processing the incoming data, which are connected to each other weighted probabilities. This name was borrowed from the experts working with the systems of artificial intelligence. Neural networks are a novelty in the foreign exchange market, and if you give a definition of a bit simplistic, the neural network - this is such a model, which in general can reproduce human brain mechanism of action and learning.

neural network model used in the field of artificial intelligence for the development of computers that could think and learn by taking as a basis the results of actions committed.

The main difference from the neural network data structure familiar to us
It is that the network take a lot of the flow of information, and the output
give one result. When it is possible to give a quantitative analysis of the data,
there is a method of adding it to, consider factors in
forecasting. Networks are very often used to make predictions on
the foreign exchange market, because they can be configured to interpret data from the
followed by obtaining the findings.

For the application of neural network in predicting the forex market at the beginning
We need to "teach" it to identify and correct patterns arising
between input and output in the market. You need to spend time not only on the
setting, but also on the training followed by testing, but then
neural networks will gain the ability of taking as the basis of historical data
given the forecast future results. Is based on the idea that when
the availability of examples of pairs of incoming and outgoing data, neural network
"Learns" dependencies and then applies them to re-enter data.
Thus, the network compares its previously made, outputs for
determine the accuracy of its forecast. At the same time, it can come back
migrate back and the weight of the various dependencies achieving
so the correct answer.

For all of the above is required to train the neural network.
This can be done with a pair of different sets of data: set
data for training and testing data set. Neural networks
They have a significant advantage - they can always continue
learn by comparing its predictions with ever
incoming data. Among other things, the network can be combined
fundamentals of technical and use them. With help
its own power, the network detects possible patterns and unrecorded
It uses them to compile the forecast and produce the output most
accurate result.

However, it should be noted that this advantage may well
be a disadvantage when used to make predictions on
the foreign exchange market. Output data are good, as well as input. They
well suited for correlation, despite the fact that you can
enter the huge flow of information at the entrance. Even under the condition that
no relationships and patterns, they are well isolated from different patterns
data types. This is a significant advantage - the ability to
Use intelligence eliminating emotions. However, there is a spoon
tar. This advantage can also be a weakness. when
volatile arises unknown factor, neural networks, the foreign exchange market
They can not give it an emotional weight.

Now in the forex market are trading platforms, which include
neural network itself, and even a special technology, which enables the
"Train" network of your trading system to make predictions, and to
their basis to generate orders for sale and purchase.

The main rule in the construction of a neural network in trading is
self-study. Know what you're doing, and all the time to expand your knowledge.
If you are going to succeed, no matter what you are working
(Analytical indicators, with technical analysis with neural networks and
et al.), you should learn as much as possible of the new.