Documentation

We take earnings-call transcripts, embed them, and score them with quant signals before you ever see them. Read it over a REST API, or query it in your own warehouse as Apache Iceberg tables. You scrape nothing and run no ETL of your own.

Two product lines share one API and one catalog. The embeddings are 768-dimensional vectors over earnings-call transcripts, chunked per fiscal period, and every vector points back to its ticker, fiscal_period, and chunk_idx, so citations come free. The signals are weekly Quant Scores: Piotroski F-Score, Altman Z-Score, Beneish M-Score, Sloan accrual, an HMM market regime, and a composite, currently across the S&P 500 (beta).

Every record is bitemporal. It carries effective_ts (when it was true) and knowledge_ts (when we learned it), which is what keeps a point-in-time backtest from seeing data it could not have known yet. We append new rows rather than updating old ones, so history never changes underneath you.

Base URL https://api.vectorfinancials.com. Start with Authentication, then jump to Examples.