STSS - Simple time series statistics



A simple yet reasonably robust tool for applying statistical analysis and quantitative transformation techniques to time domain data arrays through visual flow chart design and execution. The areas of application of STSS include (but are not limited to) technical analysis of forex and stocks, investment/hedge funds returns history analysis, simple signal filtering and time-centric experiments observations analysis.

The manual (with screenshots!) is available on the Documentation page of this site, please see the navigation bar above. Please take a look at it (it is a simple and practical introduction guiding you from the basics to what can be considered advanced parts quickly). Though STSS is designed to be as intuitive and handy as possible (no menu crawling or code writing, a document view is meant to give an immediate understanding of the whole experiment logic at a glance) there are some basic concepts (that you will need to know to get started) that are not very intuitive.

The source code is available as a zip file in the Downloads section (no code versioning system used).

The application is being implemented in C# for the WinForms (.Net 4.0) platform (meant to work on all desktop/laptop PCs running 32 or 64-bit versions of Windows XP/7/8 (though only 32-bit Windows 7 with .Net 4.5.1 is tested), may also work under GNU/Linux with Mono (theoretically, not tested, feedback is welcome)).

The project is currently at the prototype stage, the public alpha version offered for download still contains a serious amount of bugs to fix, exceptions to handle, quick-and-dirty solutions to refactor and usability and functionality improvements to implement but is already quite playable and can be used for some real-world tasks (though I don't encourage to rely on it at this stage).

Data processing functions already implemented include some sample signal processing (4-component Laguerre filter, simple moving average and Fisher transformation), statistic (standard deviation, Pearson and Spearman correlation coefficients), neural network regression/classification and investment portfolio risk/return analysis (Sharpe ratio, Sortino ratio, MPPM (Manipulation-Proof Performance Measure)) functions as well as auxiliary primitives like series shift, difference, output normalization and others. Many more are coming soon including analytical strategy backtesting, spectrum analysis (Fourier transformation), adaptive digital filters and others.

The source code is structured with maintainability and extendability in mind making it easy to add functions of your own with just a couple of edits.

The output is displayed on a chart and in a data grid (a table) which can be customized and exported to a CSV file to be used as input for other tools or included in a document.

The projects ("documents") are saved in a simple, easily readable (by a human) and parseable (by a machine) machine-independent (no local-specific data included) XML format which makes it easy to exchange your experiment settings with colleagues and to develop your own programs capable of working with the same format.

The application is released under the BSD license for free use by anyone. The application also makes use of the FlowGraph component that is available under the BSD license too (© Ivan Mashchenko and Thomas Iche). The FlowGraph source code is also included in STSS source code packages.

I am also going to provide limited free support to enthusiast users and commercial priority support to those seeking it.

May you be interested in a discussion - feel free to contact me through my profile here at CodePlex or at Google+.

Last edited Apr 23, 2014 at 12:02 PM by StrangeAttractor, version 20