Master’s Thesis · LLM Experiment on Earnings Headlines
How do framing and source credibility affect investor interpretation of earnings news?
Building an LLM experiment with GPT-5 + FinBERT (via Hugging Face) on 1,000+ synthetic headlines - contrasting social media (Reddit) with financial press (Wall Street Journal).
First LLM-based thesis in this niche. Highly relevant for large-cap firms: even subtle Big Tech earnings headlines (isolated from sentiment effects) can alter perceived credibility and caution.
Demonstrates how subtle framing shifts can change investor expectations - insights that could guide corporate communications teams and investor relations in managing market reactions.