Methodology

Understanding how we compute sentiment, detect earthquakes, and measure propagation.

Data Sources

We ingest data from SEC filings, earnings call transcripts, real-time news feeds, commodities exchanges, regulatory announcements, and select financial APIs. All sources are validated for recency and accuracy.

Sentiment Computation

We apply NLP models to extract entity mentions and emotional tone. Each node gets a rolling sentiment score (bearish / neutral / bullish) updated continuously. Scores are confidence-weighted.

Earthquake Detection

Major events (earnings surprises, regulatory decisions, market shocks) are detected via spike analysis and news clustering. We flag events with magnitude 6+ on our internal scale.

Propagation Modeling

Using graph algorithms, we simulate how a shock spreads through the network based on edge weights and node sensitivity. Results include confidence intervals and affected paths.

Important Disclaimers

Not investment advice: Nodestock is an intelligence and visualization tool only. Our sentiment scores and propagation models are probabilistic and should never be the sole basis for investment decisions.

Sentiment is probabilistic: All sentiment scores come with confidence intervals. Past sentiment does not guarantee future market behavior.

Data sources: We validate sources, but markets can be affected by events we haven't captured or modeled yet.

Always consult advisors: Combine Nodestock insights with your own research and professional financial advice.