Sentiment Analysis on FOMC Statement

Credits

This sentiment analysis was based on the ProsusAI model, which can be found at this github link: https://github.com/ProsusAI/finBERT

Installation

Follow the instructions here for installing libraries https://github.com/ProsusAI/finBERT/blob/master/README.md

Model

The dataset used for this FinBert sentiment analysis was the Financial PhraseBank from Malo et al. (2014). The dataset can be downloaded from this link.

Input

The input for this model was the FOMC statement from August 2025 titled "Statement on Longer-Run Goals and Monetary Policy Strategy", which can be found at this link. The contents of this file is in ./fomc-report.txt.

Running the code

Navigate to root folder and run the following code in the command line:

python -m scripts.predict --text_path fomc-report.txt --output_dir output/ --model_path models classifier_model/finbert-sentiment

Output

Output file is in output/prediction.csv and is described by 'negative', 'neutral', or 'positive' for each sentence.