Neural Network Interface

Artificial intelligence neural network techniques are a modern mathematical tool allowing relatively easy analysis of complex data interactions due to ability of the computer to, ineffect, program itself. A major feature of Stock Prophet is the ability to apply the neural network technique to the analysis of financial markets. A significant capability of neural networks is an ability to combine MULTIPLE indicators into a SINGLE indicator. Themathematical relationship used in this combination process is generated by the neural network computer learning process. The resulting single neural network derived indicator provides the decision variable in a trading system. This single indicator provides clear trading signals given days and weeks BEFORE execution of the trade. This is in stark contrast to essentially all other technical indicators which are late due to smoothing required to minimize whipsaw trades. This process requires relatively little input from the user and greatly enhances the prospects for rapid development of successful trading strategies.

Stock Prophet enhances the ability to develop a neural network based trading system by providing an interface capability with the BrainMaker neural network program and ability to export ASCII training files to EXCEL. Stock Prophet handles the complex task of generating the required neural network input data files thereby allowing straightforward access to neural network automatic learning capabilities. Neural network users estimate that up to 95% of the total effort needed to develop a neural network is spent in collection, organization and preprocessing of the training data. It is this 95% of the effort which is made efficient by Stock Prophet. The data can be downloaded from a commercial data base and processed by Stock Prophet for input into BrainMaker and other neural network learning programs. Stock Prophet is designed to handle the considerable detail of this process with minimum user effort.

Preprocessing of input data has been found to be a key factor for efficient learning by BrainMaker. The indicators and mathematical operations provided in Stock Prophet provide the required preprocessing options.

The fully trained neural network indicator is imported into Stock Prophet for historical testing and actual trading.

Profitability testing