log in  |  register  |  feedback?  |  help  |  web accessibility
Logo
Teaching machines to read for fun and profit
Gary Kazantsev - Bloomberg LP
Wednesday, October 30, 2013, 11:00 am-12:00 pm Calendar
  • You are subscribed to this talk through .
  • You are watching this talk through .
  • You are subscribed to this talk. (unsubscribe, watch)
  • You are watching this talk. (unwatch, subscribe)
  • You are not subscribed to this talk. (watch, subscribe)
Abstract

Does news move prices of financial instruments? The literature is divided on this question; ultimately the answer reduces to questions about validity of various versions of the efficient markets hypothesis. In this talk we describe work investigating this question empirically, driven by the objective of building predictive models of market behavior.

The objective of the specific research we describe is to build predictive models of equity prices based on text of financial news stories (and ultimately other factors). This problem is challenging in several different dimensions due to the nature of the models required: the work is at the intersection of text analysis using supervised learning and financial time series modeling via regression analysis and semi-supervised anomaly detection.

We will give an overview of the evolution of this problem at Bloomberg, describe the project as it stands and conclude with a discussion of future directions and a Q&A session.

Bio

Gary Kazantsev runs the R&D Machine Learning group at Bloomberg LP, leading projects in the areas of computational linguistics and machine learning such as sentiment analysis, market impact indicators, machine translation, text classification and predictive modeling of financial markets. He holds degrees in physics, mathematics and computer science from Boston University.

This talk is organized by Jimmy Lin