Not logged in.
Quick Search - Contribution
Contribution Details
| Type | Master's Thesis |
| Scope | Discipline-based scholarship |
| Title | Improving Factor-Based Quantitative Investing by Forecasting Company Fundamentals with Earnings Call Transcripts |
| Organization Unit | |
| Authors |
|
| Supervisors |
|
| Language |
|
| Institution | University of Zurich |
| Faculty | Faculty of Business, Economics and Informatics |
| Date | 2021 |
| Abstract Text | Factor investing with lookahead fundamental features significantly outperforms those with historical fundamental features. Motivated by this insight and the great amount of unstructured textual data about companies, we use textual features extracted from earnings call transcripts by a pre-trained FinBERT model and financial features to predict year-ahead earnings. Additionally, we construct risk-adjusted portfolios incorporating uncertainty estimates, which significantly outperform standard factor models. Specifically, our portfolio based on the RNN uncertainty-aware model with FinBERT textual features has an annualized return of 15.39% (vs. 9.43% for S&P 500 total return) and a Sharpe ratio of 1.29 (vs. 0.71). This result is robust to the control of the Fama-French five factors, and our portfolio shows a significant positive alpha of 0.32% per month. |
| Export |
BibTeX
|