Stock Data with semantic context

This dataset provides a comprehensive, multi-modal resource for financial time-series analysis and stock prediction models. It integrates daily historical stock prices with two critical streams of contextual information.The first stream consists of daily aggregated news sentiment scores, designed to capture real-time market perception. To ensure data continuity, days where new sentiment data could not be found inherit the score from the previous day, to which a decay function is applied.The second stream incorporates quantitative features derived from periodic corporate disclosures, specifically quarterly financial sheets and conference call transcripts. A novel feature of this dataset is the temporal modeling of these disclosures: key analytical values are introduced on their exact release date, and a separate decay function is subsequently applied to these values for all following days. This methodology models the diminishing informational relevance of an event, which persists until the next disclosure event resets the values.This dataset is ideally suited for developing and testing predictive models that combine continuous market data with event-driven, fundamental analysis.