the concept of neural networks is rapidly increasing in popularity in the area of developing trading systems. at one point in time, it would have seemed impossible to make a system that would be able to adapt to changing markets, but recent developments in technology have now made having these types of systems a reality.
as far as trading is concerned, neural networks are a unique method of technical analysis, intended for traders who take a thinking and strategic approach to their business. for a serious, thinking trader, neural networks are an excellent tool that can detect subtle non-linear interdependencies and patterns that other methods of technical analysis are unable to uncover.
well-prepared input information on the targeted indicator is the most important component of successful trading with neural networks. neural networks and the resulting trading system can only be as good as the dbases provided.
to trade audnzd successfully we pay close attention to all the stages of the network preparation cycle. it's our business to invent ideas, formalize, test and improve them, and, finally, choose the right moment to dispose of them. every neural network based model has a life span and cannot be used indefinitely.
the longevity of a models life span depends on the market situation and on how long the market interdependencies reflected in it remain topical. we either retrain the model using completely new data, add some new data lines to the existing data set and train the model again, or simply retire the model altogether.
data is derived from from a wide scale of technical indicators derived from the audnzd quote itself as well as from development of its influence factors. we basically extract every bit of useful data the financial markets supply.
selection of input factors, adjustment of parameters, training and testing of the neural models is a continous process taking up a lot of time and requiring years of experience and expertise. each neural network which we employ to forecasts aussie/kiwi can be viewed as an expert in itself. the combined "opinion" of these experts is the daily signal which we send out to our clients.
our neural nets are used as a committee, which is another benefit of our strategy. each of our multiple nets can be responsible for some specific aspect of the market, giving us a major advantage all across the board.
neural network technology is an analysis method used to identify underlying relationships, structures and patterns in a set of data by using a process that mimics the way the human brain operates. neural networks have the ability to learn and adapt to changing input.
neural networks emulate certain major aspects in the functioning of the human brain. this gives them a unique, self-training ability, the ability to formalize unclassified information and, most importantly, the ability to make forecasts based on historical information. neural networks bridge the gap between human intelligence and computer systems.
download an excellent introduction to neural networks by Kevin Gurney.
for absolute neural network junkies among you fx traders out there we recommend the following download by simon haykin: neural networks & learning machines