Mining for Patterns in Financial Time Series

• Activity Number: 490
• Type: Topic Contributed
• Date/Time: Wednesday, August 1, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): James Gentle*+
• Companies: George Mason University
• Keywords: clustering of time series subsequences ; streaming data ; rule discovery

Abstract

There is a widespread belief that certain patterns of stock prices over time portend specific future types of movement of those prices. We consider various criteria for identifying these types of patterns, and briefly look at some historical price data. We then look at various specific types of breakout movements, and then attempt to determine bags of patterns that tend to precede these breakouts. In the absence of concomitant data, patterns within a single time series of stock prices seem to have very little predictive power. We consider types of patterns over multivariate time series, methods of identifying such patterns, and whether multivariate patterns have more predictive power.