Research & Risk Disclosures

Read this before using any score, ranking, model output or performance figure. These disclosures are part of every research publication on this site.

Scores and rankings are probabilistic research outputs, not facts or trade instructions. Large language models can hallucinate, misread sources, reproduce bias, omit material information and change behaviour between versions. Public filings, transcripts, news and market data can be incomplete, delayed, revised or wrong. Human review reduces but does not eliminate these risks.