Stock Seeker: Finding Correspondences in Stock Data∗

The problem of stock data analysis has been widely researched by stock analysts interested in detecting patterns in stocks. Resulting algorithms have dealt with many representations of stock data depending on the problem, many dealing with complex cases where patterns in stock data need to be matched despite differences in scale, translation, or noisiness. However, specific instances arise where such generality is not needed. For example, analysts may be interested in compiling historical price data from different sources. In this case the correspondence between subsequences in two datasets is not known, but is desired to allow analysts to have accurate stock prices. This problem may be subject to noise, but is not subject to scaling or translation, allowing for a much simpler matching algorithm that can be made to run rapidly to allow for user interaction. This paper presents a visual interface that allows the user to solve such matching problems by iteratively making queries, viewing results, and changing parameters or matching algorithms.