Enhancing Brazilian Inflation Forecasts through Sentiment Analysis Using Large Language Models

2026-04-12 · Source: Paper Index on ACL Anthology · Field: Finance & Economics — Economic Analysis & Policy, Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Expert, quick

Summary

This study introduces a framework to enhance Brazilian inflation forecasts by integrating sentiment analysis derived from Large Language Models (LLMs). The framework extracts sentiment variables from the Brazilian Monetary Policy Committee (COPOM) minutes, optimizes these variables to align with human-collected sentiment, and then incorporates them into ARIMA and LSTM models for one-step-ahead monthly IPCA (consumer price index) prediction. The research found that LLM-generated sentiment trends are temporally coherent with historical inflation patterns and demonstrate high statistical significance (p < 0.001). Models using sentiment evaluations that closely matched human assessments, such as grok-4-fast and llama-4-maverick, achieved superior forecasting performance. ARIMA models consistently improved with sentiment inclusion, whereas LSTM model results showed more variability.

Key takeaway

For NLP Engineers developing economic forecasting models, incorporating LLM-derived sentiment analysis can significantly improve predictive accuracy. You should focus on optimizing LLM sentiment outputs to align with human expert assessments, as this alignment directly correlates with superior forecasting performance. Consider integrating these sentiment variables into traditional econometric models like ARIMA for more consistent gains in inflation prediction.

Key insights

LLM-derived sentiment from policy minutes significantly enhances inflation forecasting accuracy when aligned with human assessments.

Principles

• Qualitative signals improve econometric models.
• LLM sentiment can be optimized to human judgment.

Method

Extract sentiment from policy minutes using LLMs, optimize LLM bias to match human sentiment, then integrate into ARIMA/LSTM for one-step-ahead monthly inflation prediction.

In practice

• Use grok-4-fast or llama-4-maverick for sentiment.
• Integrate sentiment into ARIMA models for stability.

Topics

• Brazilian Inflation Forecasts
• Sentiment Analysis
• Large Language Models
• Monetary Policy Committee
• ARIMA Models

Best for: NLP Engineer, AI Scientist, Research Scientist, Data Scientist

Related on AIssential

• Beyond Polarity: Multi-Dimensional LLM Sentiment Signals for WTI Crude Oil Futures Return Prediction — Multi-dimensional LLM sentiment, beyond polarity, improves WTI crude oil futures return prediction.
• Transformer-based CoVaR: Systemic Risk in Textual Information — Integrating raw LLM text embeddings with market data improves systemic financial risk (CoVaR) estimation.
• Introduction to quantitative finance Part 26: Whether to use forecasting methods, or to tell an LLM… — Forecasting in quantitative finance faces challenges with LLMs, necessitating memory-enhanced reward shaping for consistency.
• What sentiment analysis can't see: Measuring whether customers were helped, and what went wrong, across 70,000 support conversations — LLM-based analysis of customer support conversations provides deeper insights into satisfaction and underlying problems than traditional sentiment analysis.
• A Larger Annotated Corpus of Portuguese Coreference — LLMs can annotate Portuguese coreference, but a new LLM-based system did not outperform existing rule-based methods.
• Consistency Analysis of Sentiment Predictions using Syntactic & Semantic Context Assessment Summarization (SSAS) — SSAS improves LLM sentiment prediction consistency by pre-processing noisy data into structured, high-signal prompts.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Paper Index on ACL Anthology.