Παρουσίαση/Προβολή

(OIK501) -  ΣΤΥΛΙΑΝΟΣ ΑΡΒΑΝΙΤΗΣ

Περιγραφή Μαθήματος

 

This module introduces students to modern econometric uses of financial text through the construction of functional sentiment representations and their application to market forecasting. Building on prior training in NLP and FinBERT, this pair of  lectures develop the Sentiment Utopia Index (SUI), a corpus-normalized measure of narrative extremeness that exploits the rhetorical structure of documents rather than simple document-level averages. Students will learn how to transform SEC filings into daily sentiment profiles, construct upper and lower sentiment envelopes, implement alternative weighting schemes, and integrate the resulting indices into HAR models for realized volatility forecasting. Emphasis is placed on rigorous real-time evaluation—Diebold–Mariano and Clark–West tests, HAC inference, and reproducible Python pipelines with cached intermediate outputs. The module combines methodological foundations with hands-on code walkthroughs and prepares students to extend the framework in a short research-style mini-project.

Ημερομηνία δημιουργίας

Δευτέρα, 19 Ιανουαρίου 2026