QMI members work together to answer interesting, open-ended finance questions using a wide array of quantitative tools. Projects range from analyzing the key drivers of individual equity performance to predicting broader industry trends to drawing conclusions on more general financial concepts. The questions our members attempt to answer are challenging and integral to how financial markets operate. A quick overview of this semester’s projects signify at least that much.

Current Projects

Analyzing Complexity and Sentiment of Earnings Calls Using NLP

Posted August 25, 2019
By Jessica Huang, Wesley Klock, Zander van Geenen

Earnings calls provide powerful insight into company performance, future plans, and analyst opinions. This project attempts to use Natural Language Processing (NLP) techniques to simplify and improve manual earnings call analysis. Earnings call transcripts will be collected from the top 10 market-cap companies from 5 sectors. For each company, transcripts will be analyzed going back 5-years for sentiment and complexity. These metrics will be determined using a custom earnings call dictionary and sequence based machine learning model. After these metrics are constructed and backtested, they will be regressed against traditional financial metrics to determine their viability as a leading indicator of performance. Lastly, these metrics will be used alongside company properties to cluster stocks and construct potentially viable strategies. This method of analysis aims to improve on current NLP techniques by constructing a unique dictionary to determine complexity and sentiment and tying insights on these metrics to actual financial performance.

Select Datasets

Through the course of working on their projects, our student collect interesting and unique datasets. We offer these datasets to partners and other student organizations based on request.