Shiraz Top
Top Picks









  Shiraz TrendFinder Screen Shots

The following picture shows the TrendFinder tool available to automate the model building process.  In the dialog box there are functional sections for each of the steps to import, synthesize, optimize data models and provide the final predicted trend.  For each step a set of goals is available to customize or leave default to discover data trends and prediction accuracy.  Parameters are saved as part of the document file so model updating operates exactly like the initial model build environment.



The following picture shows a typical model document containing specific parameters in addition to the historical training/testing data.  This page is implemented as a standard grid tool allowing direct import or export to such other spreadsheet applications as MS Excel or MATLAB.  The Shiraz application contains a self standing graphics display which holds the data calculated for the last training or prediction operation.  Shiraz also supports direct import and sorting of TAB delimitated or CSV type datafile commonally found in server download environments.

Each document can contain multiple (up to three maximum) symbol data for usage in building of trend analysis database.  For example if a mutual fund was broken up into several holdings, the two largest stock holding portions of the fund could be included seperately.  The fund overall symbol could be used in addition to the other two stocks.  Symbol data is stored left to right with the extreme right portion (beyond the imported data) reserved for custom user entered data such as dividend payouts or other market conditions.




The following picture shows usage of simple to interprete figures of merit for model training and prediction quality.  All portions of the Shiraz GUI have been designed to eliminate most references to the underlying math details.



The following picture details the Shiraz WatchList document which allow the manipulation and usage of multiple model documents in a protfolio type arrangement.  Each entry consists of a complete model document with embedded data and model parameters which can be updated seperately or in a batch mode.  After a specific entry is updated the date column contains the a timestamp and related status in the form of a red/green colorization.  Watchlist documents can be saved/opened like model documents and handling of the list entries is automated to ease user manual manipulation of files.