DJIA Historical Market and Financial News Sentiment Dataset for Stock Price Forecasting

This dataset contains historical Dow Jones Industrial Average (DJIA) market observations and associated financial-news information used for next-day stock closing-price forecasting. The dataset supports the research study on multimodal financial time-series forecasting using adaptive multi-view fusion and dynamic hypergraph learning. The dataset consists of two complementary data sources. The first file, DJIA_2000_2016.csv, contains daily DJIA market observations, including date, opening price, highest price, lowest price, closing price, and trading volume. The second file, stock_senti_analysis.csv, contains daily news information associated with the DJIA, including a binary label and up to 25 news headlines per trading day. The market data cover the period from January 3, 2000, to July 1, 2016. The news dataset covers the same overall period and provides daily textual information associated with the corresponding trading dates. The two sources are aligned by trading date for subsequent multimodal financial forecasting. In the associated study, the financial-news information is processed using FinBERT to obtain sentiment-related features, including aggregated negative sentiment, positive sentiment, polarity, sentiment intensity, and rolling sentiment representation. Technical market features are additionally derived from the historical market variables, including daily returns, lagged returns, high-low price differences, and open-close price differences. The resulting numerical, technical, and news-sentiment representations are used to construct synchronized historical windows for next-day stock price forecasting. The final modelling dataset contains 4,098 chronological observations after the data preparation and alignment procedures described in the associated manuscript. This dataset is provided to support transparency, reproducibility, and further research in multimodal financial time-series forecasting, sentiment-aware stock prediction, and dynamic hypergraph-based financial modelling.