Exploring financial correlation clusters allows investors to avoid undertaking excessive risks and discover investment alternatives. Strongly correlated assets suggest a tendency for them to gravitate towards a similar set of economic factors. Therefore, a plethora of research has focused on developing technical features to capture the dynamic comovement of stock prices. However, this analytical task faces substantial challenges arising from a large number of pairwise comparisons, the dynamic nature of correlation, and the ambiguity in understanding implication. When making investment decisions, investors also apply their knowledge of business relationships for extrapolation purposes, albeit promising technical analysis. In this work, we propose Prismatic, a visual analytics system that integrates historical performance analysis and business knowledge graph exploration to interactively cluster the dynamic financial correlations. Prismatic facilitates three key analytical processes: cluster exploration by holistically overviewing the changing structure of clusters, cluster verification by shedding lights on the temporal patterns in correlations at different time scales, and cluster generation by embedding the business knowledge graph to contextualize the underlying relationships. We evaluate the usefulness and effectiveness of Prismatic through two case studies and extensive interviews with domain experts.
The client services are provided by Vue 3. All codes below should be run under the "client" folder.
The packages are handled by npm, and specified in the package.json.
$ npm install
The client is served at http://localhost:8080/. Notice that the framework supports hot-reloads, so that the changes on DOM are applied automatically and do not require reloads.
# Compiles and hot-reloads for development
$ npm run serve
# Compiles and minifies for production
$ npm run build