In the market today there are numerous Forex signal services. One extremely difficult question to answer is, "which services are legitimate, responsible and profitable?" One strategy for differentiating one signal service from another is to examine the technology employed by the agent for trade selection. To provide the greatest opportunities for gain, automated Forex signal services monitor the markets 24 hours a day and can greatly increase exposure to profitable trades than standard, manual human trading. This does, however, lend itself to being exposed to the rigid analysis performed by the trading algorithm vs. having the flexibility of experience and intuition possessed by a human trader.
Obviously, the perfect solution would be too combined the tireless availability and rigid behavior of a Forex robot with the experience and intuition of a successful human trader. At the forefront of trading algorithm design, this is exactly the scheme that is in process today. The newest designs of algorithmic trading which employ AI are focusing on using these machine learning technologies to teach trading systems to trade more like a human being rather than the previous strategies of trying to use these mechanisms to forecast market direction.
One current and successful application of this strategy employs a machine learning technology called genetic programming or GP. Genetic programming is a computer programming technique in which the Genetic Programming application actually scripts programs. These programs are rather simple. They are a group of nodes and branches in a tree structure contracting of operators (+, -, *, /, etc) and variables. These programs are then tested on a dataset and their outputs are measured against a desired result to determine the program's fitness. The most fit programs are selected to continue. These programs are then crossed with other programs from the population generated by the GP application. The process begins again; the most fit programs are selected, are again crossed and retested. The entire process continues until the GP application has found a program meets a minimum fitness standard. The final program becomes the model algorithm of the data set.
Until recently, most applications of these technologies in the world finance were limited to processing data sets in an attempt to forecast a market direction a given number of periods into the future. The strategy has met with only limited success as the modeling created through machine learning is very pattern oriented, and bids to over-fit data sets created by financial time-series. This produces results which are not robust and fail when applied to fresh market data.
There are some new views on using machine learning, GP specifically, to create trading algorithms that are humanized. This is being completed by substituting a human decision data set for a financial time-series data set. This creates some fundamental differences in the outcomes of trading systems use these algorithms. The old way attempted to learn patterns from a near random financial time-series. This proved to be very difficult and many patterns which were recognized by the machine learning technologies do not commonly repeat. This new method learns patterns from the behaviors of a human being, a successful human trader. Humans are creatures of habit; their patterns of habit are far from rare and are easily recognizable using GP and the appropriate fitness function.
This remarkable discovery is producing new algorithms that enhance the basic strategies employed by automated trading robots. By letting the basic robot rules determine potential trade setups, then employing the GP designed human algorithm to filter a potential setup and decide if the opportunity valid, these robots become more selective in their trading and more profitable in their results.
Figures contend that of all human traders, only a few percent make consistent profits the long-term. Trading robots perform far worse. Through combining the humanized algorithm with automated trading rules, trading robots will continue to close the gap on human traders.
