650 – Miscellaneous Computing
Identifying Patterns in Financial Time Series
James Shine
George Mason University
James E. Gentle
George Mason University
Charles Perry
USDA
One important application in financial and other time series is the construction of trendlines (both positive and negative) for specific time periods. Such trendlines have been done visually and manually for decades, but computational methods for automatic or semi-automatic trendline generation is a much newer science. We have developed an algorithm for generating trendlines from specific time series data observations using minimum values, moving averages and constrained optimization. We will show results of this algorithm on various sets of time series data. This tool could prove useful for future analysis of finanacial and other time series.