Trillion Dollar Words in Julia - Natural Language Processing for Financial Markets

Learn how to predict the impact of Federal Open Market Committee (FOMC) communications on financial markets through a conference talk that explores the TrillionDollarWords.jl package in Julia. Discover how to work with a dataset of 40,000 time-stamped FOMC sentences spanning from 1996 to 2022, and leverage a RoBERTa model with 355 million parameters to classify statements as 'hawkish', 'dovish', or 'neutral'. Master loading and manipulating financial datasets, implementing model inference, and analyzing layer-wise transformer embeddings for market indicator predictions. Explore the intersection of Economics, Finance, and Artificial Intelligence while learning to fine-tune models for various classification tasks using Transformers.jl. Gain hands-on experience with practical examples, including accessing layer-wise activations and evaluating model performance through root mean squared error analysis.