Stock Sentiment Analysis

Fetch recent news for a ticker and compute a sentiment score using the OpenAI API.

Setup

1. Create a local .env file (optional) in the project root:

OPENAI_API_KEY=...
# Optional (only needed for --source newsapi; otherwise Google News RSS is used)
NEWSAPI_KEY=...
# Optional
OPENAI_MODEL=gpt-4o-mini
OPENAI_BASE_URL=https://api.openai.com/v1

2. (Optional) Install a CLI entrypoint:

python3 -m pip install -e .

3. Run an analysis:

python3 -m stock_sentiment analyze TSLA
python3 -m stock_sentiment analyze TSLA --days 7 --max-articles 50 --format json --include-reasons
python3 -m stock_sentiment analyze TSLA --source google-rss

Notes:

• Default output is a single-line text summary; use --format json for structured output.
• By default --source auto uses NewsAPI when NEWSAPI_KEY is set, otherwise Google News RSS.
• In --source auto, if NewsAPI fails the CLI falls back to Google News RSS.
• OPENAI_API_KEY is required unless all needed per-article classifications are already cached.
• Add --include-articles to embed article metadata in JSON output.
• Add --verbose to print per-article sentiment details in text mode.
• OpenAI results are cached locally by default (see --cache-dir, --no-cache, --cache-ttl-hours).
• JSON output includes source and lookback_days fields for downstream systems.

Disclaimer: This tool is for informational purposes only and is not financial advice.

Tests

python3 -m unittest discover -s tests -p "test_*.py"