How Stocktiment
works
Stocktiment aggregates financial news from multiple sources and uses large language models to classify sentiment in real time — producing a Fear/Greed score for each tracked asset.
Where the data comes from
For each ticker, up to 15 recent articles are collected from three independent sources. Only articles published within the last 7 days are used — older content is discarded to ensure the score reflects current market sentiment.
How the score is calculated
Each article headline is classified as positive, negative, or neutral by a large language model (Llama 3.1 via Cerebras or OpenRouter). The raw score is then computed as a weighted average:
When data is refreshed
The 75 tracked tickers are split into three lots of 25, scanned in rotation every 8 hours. Each lot is scanned once per day, ensuring all tickers are refreshed within a 24-hour window without overwhelming the sentiment API.
S&P 500 · NASDAQ
CAC 40 · DAX
FTSE · China ADR
Custom ticker searches (via the search bar) are analyzed on demand and are not included in the scheduled scan cycle. Their results are not persisted between sessions.
Sentiment scores are derived from automated NLP analysis of publicly available news headlines. They do not account for fundamental analysis, earnings data, macroeconomic conditions, or any other factors relevant to investment decisions.
Always conduct your own research and consult a qualified financial advisor before making any investment decision. Past sentiment patterns are not indicative of future price movements.