[Back to projects](https://yudis.dev/en/portfolio) ![Screenshot of News Sentiment & Bond Yields](/assets/sun-scraping-6noFvCBi.webp)

# News Sentiment & Bond Yields

Python Scraping NLP Sentiment Analysis 2024

Research with ITS and Indonesia’s Ministry of Finance on financial-news sentiment and bond yields.

Closed research project

News Sentiment Analysis is a collaborative research project conducted jointly by Institut Teknologi Sepuluh Nopember (ITS) and the Ministry of Finance of Indonesia, investigating the relationship between news sentiment and Indonesian government bond market yields.

## Research Objective

The project examines whether sentiment extracted from Indonesian financial and economic news articles has a measurable impact on bond yield movements. By quantifying the "mood" of news coverage, the research aims to provide a data-driven signal that supplements traditional bond market analysis.

## Key Components

### 1\. Automated News Scraping

A Python-based scraper collects articles from major Indonesian financial and economic news sources on a scheduled basis. The pipeline handles pagination, deduplication, and raw HTML extraction, storing structured article data including headline, body, publication date, and source.

### 2\. Text Preprocessing

Raw article text is cleaned and normalized for NLP processing: removing stopwords, handling Indonesian-language tokenization, and preparing inputs for sentiment classification models.

### 3\. Sentiment Classification

Each article is scored using NLP models adapted for the Indonesian language and financial domain. Sentiment scores are aggregated by day and news source to produce composite sentiment indices.

### 4\. Bond Yield Correlation Analysis

Sentiment indices are compared against daily Indonesian government bond yield data across multiple maturities. Statistical analysis — including time-lag correlation and regression modelling — tests whether sentiment leads, lags, or moves concurrently with yield changes.

## Purpose and Benefits

The research contributes to the understanding of how public information and media narrative influence sovereign debt markets in an emerging economy context. Findings support more informed bond market analysis by incorporating textual data alongside traditional macroeconomic indicators.

## Have a similar challenge?

Tell me what you’re building and where you’re stuck.

 [Get in touch](https://yudis.dev/en/#contacts)

---

Canonical: https://yudis.dev/en/portfolio/news-sentiment-analysis