Quantifying market mood from tweet storms and headlines

What it does

stocksight ingests tweets and news headlines for a given stock symbol, runs the text through Python sentiment analyzers like VADER and TextBlob, and stores the scored emotions in Elasticsearch. A companion script pulls raw stock prices into the same index, letting you overlay market data against the mood of financial Twitter. You can filter the firehose with NLTK token whitelists and blacklists to ignore noise.

The interesting bit

Rather than stopping at 280 characters, the scraper can follow URLs embedded in tweets and headlines to read and score the full landing-page text, weighing the article behind the clickbait.

Key highlights

• Mines live Twitter streams or news headlines by stock symbol and keyword.
• Applies multiple sentiment analyzers—VADER, TextBlob, and NLTK—to incoming text.
• Recursively scrapes linked pages to score the full story, not just the snippet.
• Stores time-series sentiment and price data in Elasticsearch for Kibana dashboards.
• Orchestrates the scraper, Elasticsearch, and Kibana via Docker Compose.

Caveats

• The README targets Elasticsearch 5.x and Kibana 5.x, which are several major versions behind current releases.
• The authors explicitly label this educational software, warn of potential bugs, and disclaim any responsibility for trading results.

Verdict

A solid starting point for data engineers or quant-curious developers who want to correlate social sentiment with market data without building a pipeline from scratch. Skip it if you need production-grade financial tooling or a modern Elasticsearch stack.