1.1. What is NeuroForecaster?

NeuroForecaster is an advanced windows-based, user-friendly business forecasting tool. It is packed with the latest technologies including neural network, genetic algorithm, fuzzy computing and non-linear dynamics. It excels at:

Time-Series Forecasting

eg. Stock & Currency Market Forecasts, GDP Forecast

Classification

eg. Stock Selection, Bond Rating, Credit Assignment, Property Valuation

Indicator Analysis

Identification of useful input indicators

Fig.1. The NeuroForecaster main screen with the QuickStep buttons.

With NeuroForecaster, you can create neural networks of any size - the only limit is the memory constraint of your computer. It is also an excellent tool for analyzing the effectiveness of neural network architecture (numbers of hidden nodes, hidden layers, transfer functions, etc) for problem solving. It accepts numerical input attributes, patterns, codes, technical indicators and fundamental indicators, and allows them to be combined to build uni-variate or multi-variate models. Like other numerical tools, it cannot handle descriptive information such as news, rumors or fiscal policies unless such information is accompanied by numerical information, reflected in other indicators or can be quantified.

It is a general-purpose forecasting tool with adaptive learning capability, you can make it into an auto trading system if you train it with a good trading strategy to generate the buy/sell signals.

1.2. What is GENETICA Net Builder?

GENETICA Net Builder is supported by NeuroForecaster 3.0 and above. It searches for the best combination of input data, the most predictable forecaster horizon (i.e., how many steps ahead to forecast), and automatically creates and optimizes forecaster structures and control parameters by evolution and genetic search.

Fig. 2. The evolving network population.

GENETICA is seamlessly integrated with the training process of NeuroForecaster -- it determines when to pause the training, purge the poor-performing networks and evolve new offspring to inherit the knowledge acquired by outperforming parents.

1.3. Powerful and Unique Features

• Reads various file formats including MetaStock, CSI, Computrac, FutureSource, ASCII and VisuaData, displayed in its built-in spreadsheet.

• GENETICA Net Builder which searches for the best input data combination and forecast horizon, and builds the best forecaster networks
  
  
  Fig. 3. The input combination and forecast horizon searched by GA.

• 12 neural network models including Backprop, Fastprop, Radial Basis Function, NeuroFuzzy
  
  Fig. 4. The 12 network paradigms and the GENETICA Net Builder.

• Rescaled range analysis & Hurst exponent to unveil hidden market cycles & check for predictability
  
  Fig. 5. Rescaled range analysis to determine the hidden cycles.

• Correlation Analysis to compute correlation factors to analyze the significance of input indicators
  
  Fig. 6. Correlation analysis to identify leading indicators.

• Weight and AEI (Accumulated Error Index) Histograms to monitor learning process
  
  Fig. 7. The learning process with the histograms.

• Point-and-click, color graphical user interface
  
  Fig. 8. The main chart with a user-friendly interface. The forecast is shown in blue, the target in red, and the training range as dotted line.

1.4. Financial, Business & Engineering Applications

These applications are included in the package:

• Stock price prediction
• Stock market 6 monthly returns forecast
• Stock selection based on company ratios
• US$ to Deutschmark exchange rate forecast
• US$ to SGD exchange rate forecast
• US$ to Yen exchange rate forecastCredit rating of bank loan applications
• Property price valuation
• Construction demand forecast
• Soccer Match Prediction
• XOR - A classical problem
• Chaos - Mackey-Glass chaotic time series prediction
• SineWave - To demonstrate Rescaled Range analysis

Based on those examples above, you will be able to create your own forecasters for many other applications such as:

• Stock market index forecast
• Interest rate forecast
• Commodity trading
• Bond rating
• National GDP forecast
• Sales forecast Fraud detection
• Fault diagnosis
• Business site selection
• Rain-fall & catchment run-off prediction
• Air passenger arrival forecast
• Recruitment screening
• Selection of insurance policies

1.5. System Requirements

IBM PC 386/486 with mouse, or compatibles, math coprocessor recommended. MS Windows 3.1, MS DOS 5.0 or above, 3.5" floppy disk drive, 4 MB RAM, 5 MB available harddisk space minimum. VGA color monitor or better.

Papers Available:

Copyright © 1993-96, NIBS Pte Ltd.