- Cyclicity Analysis of Time-Series
- Anomaly Detection and Predictive Analytics of Time-Series
My current research is in the Cyclicity Analysis of Time-Series. Cyclicity Analysis is a type of data pattern recognition technique that extracts and quantifies pairwise leader follower relationships within a collection of time-series and determines a general ordering of such time-series relative to when they undergo their temporal patterns.
A current application of this technique is the analysis of stock and cryptocurrency prices. Such prices are notoriously noisy, but using Cyclicity Analysis, we can still find overall market trends that are not visible to the human eye. Please see our research results in this Medium article or in this Github repository. We would like to thank polygon.io for providing us stocks and crypto data to do this research. Polygon.io stocks API enables instant access to financial data with a developer centric, API first approach. It provides the most accurate real-time and historical data available and institutional level market data access to individual developers through their robust RESTful and Websocket APIs. We were interested in using PolygonIO's service because of their easy to use APIs. The documentation provided by PolygonIO makes it easy for anyone to get started with limited pre-existing knowledge. Their python RESTful API library allowed me to get straight into creating programs without making my own API handler. The quality and quantity of the PolygonIO data allows for us to create stock tracking and trading programs which are reliable and successful.
Kaushik, Vivek, and Daniele Ritelli. "Evaluation of Harmonic Sums with Integrals." American Mathematical Society: Quarterly of Applied Mathematics, vol. 76, no. 3, 2018, p. 577 - 600.