Uber (UBER)

Creating a reliable sentiment analysis model requires more than just measuring positive or negative language. The methodology behind Marvin Labs’ sentiment feature is designed to reflect how companies communicate with investors—grounded in primary, regulated disclosures rather than third-party noise.

Focus on Primary, High-Quality Sources

The model includes only sources where company management is under regulatory obligation to provide fair and balanced communication. These include:

• Annual and quarterly filings (e.g., 10-Ks, 10-Qs)
• Earnings call transcripts, including both prepared remarks and Q&A
• Regulatory press releases and stock exchange filings
• Select high quality conferences such as JPM Healthcare Conference, Morgan Stanley Technology Conference, and Goldman Sachs Technology and Internet Conference, and several others
• Investor days and major public presentations (e.g., WWDC, GTC)

Unregulated marketing materials—such as blogs, promotional emails, or standalone press releases—are excluded unless filed with a regulatory body.

Exclusion of Third-Party Commentary

To preserve signal quality, the model excludes all third-party sources such as:

• Financial media (e.g., Bloomberg, Financial Times)
• Social media platforms (e.g., Twitter, Reddit)

This avoids conflating company sentiment with public sentiment, which is unevenly distributed and often biased toward a small subset of highly visible firms.

Company-Level Sentiment, Not Document-Level

Rather than scoring individual documents, sentiment is assessed at the company level. The system evaluates all recent communications, weighing each by relevance, importance, and temporal proximity. This approach reflects the broader narrative companies present over time and gives greater weight to significant disclosures like annual reports.

A single sentiment score is published per company, refreshed shortly after each new disclosure (typically within 10–30 minutes).

Normalization Across Industries

Certain industries naturally use more emotive or positive language than others. To ensure fair comparison, sentiment scores are normalized:

• Relative to peers within the same industry
• Across sectors and market conditions

All scores are scaled from 0 to 100, enabling apples-to-apples comparisons.

This methodology ensures that sentiment reflects the tone and trajectory of a company’s own communications—without interference from external noise or superficial document-level scoring.

Check out the launch blog post for a deeper dive into the methodological considerations driving the Marvin Labs sentiment model.