bullbear.ninja

Sources

Data sources & credits

Data on U.S. and Korean stocks, crypto assets, and Korean laws and regulations can be traced back to its source data or connected provider. Original-source links are provided on each filing and earnings item via the Source ↗ link.

This page matters because the product is built to let readers move from summary back to source. If a claim cannot be traced to a filing, official dataset, or clearly labeled provider, it should not stay in the product.

Primary filings and issuer data

SEC EDGAR, DART, and FSC/KRX are the first layer. These are the pages we rely on for filings, structured financials, listed-security metadata, and company basics.

Market and holdings coverage

KRX, KOFIA, Yahoo Finance, and public institutional datasets provide the market context around prices, funds, and 13F-based ownership changes.

Translation and summarization stack

Google Gemini is used for bilingual translation and analysis support, while the product keeps source links visible so users can verify the original material directly.

Financial analysis & AI insight quality assurance

The company financial analysis and AI insights in this service go through a three-stage verification framework for high reliability. First, through source-data traceability, we use the originals from DART and SEC EDGAR rather than processed secondary materials, and we require original-filing links so users can cross-check them directly. Second, we apply a dedicated AI quality-assurance (QA) layer that evaluates whether each generated briefing meets internal standards, and when it falls short, we continuously improve the model through retries and bad-case logging. Third, through macro-indicator cross-referencing, we combine authoritative market data such as FRED and KRX to prevent bias in single-company analysis.

We do not store or redistribute the full text of external reports without authorization. Originals are provided as source links, and translations of company descriptions and analysis in this service (English→Korean for U.S. companies, Korean→English for Korean companies) are processed with Google gemini-2.5-flash and are labeled with the relevant sources.

Voice generation is processed with active use of OpenAI Privacy Filter and Codex. The AI models applied may differ depending on your plan.

Related pages

Use these public pages to move from source transparency into actual market, report, and product context.

We use cookies and similar technologies for analytics, to understand how the site is used. You can accept or decline non-essential cookies. See our Privacy Policy.