Guido Germano


2025

pdf bib
Leveraging LLM-based sentiment analysis for portfolio optimization with proximal policy optimization
Kemal Kirtac | Guido Germano
Proceedings of the 1st Workshop for Research on Agent Language Models (REALM 2025)

Reinforcement learning (RL) offers adaptive solutions to portfolio optimization, yet standard methods such as proximal policy optimization (PPO) rely exclusively on historical price data and overlook the impact of investor sentiment. We introduce sentiment-augmented PPO (SAPPO), a reinforcement learning framework that incorporates real-time sentiment signals extracted from Refinitiv financial news. Daily sentiment scores are generated using LLaMA 3.3. SAPPO integrates these signals into the PPO advantage function via a sentiment-weighted term, enabling allocation strategies that respond to both price movements and market sentiment. Experiments on a three-asset portfolio demonstrate that SAPPO increases the Sharpe ratio from 1.55 to 1.90 and reduces drawdowns relative to PPO. The optimal configuration uses a sentiment influence parameter 𝜆 = 0.1, as validated through ablation studies and statistically significant t-tests (p < 0.001). These findings show that sentiment-aware reinforcement learning improves trading performance and offers a robust alternative to purely price-based strategies.

2024

pdf bib
Enhanced Financial Sentiment Analysis and Trading Strategy Development Using Large Language Models
Kemal Kirtac | Guido Germano
Proceedings of the 14th Workshop on Computational Approaches to Subjectivity, Sentiment, & Social Media Analysis

This study examines a novel methodology for enhanced financial sentiment analysis and trading strategy development using large language models (LLMs) such as OPT, BERT, FinBERT, LLAMA 3, and RoBERTa. Utilizing a dataset of 965,375 U.S. financial news articles from 2010 to 2023, our research demonstrates that the GPT-3-based OPT significantly outperforms other models, achieving a prediction accuracy of 74.4% for stock market returns. Our findings reveal that the advanced capabilities of LLMs, particularly OPT, surpass traditional sentiment analysis methods such as the Loughran-McDonald dictionary model in predicting and explaining stock returns. For instance, a self-financing strategy based on OPT scores achieves a Sharpe ratio of 3.05 over our sample period, compared to a Sharpe ratio of 1.23 for the strategy based on the dictionary model. This study highlights the superior performance of LLMs in financial sentiment analysis, encouraging further research into integrating artificial intelligence and LLMs in financial markets.