"Stock price dataset: NASDAQ, NYSE, TSE, & AHK"

"This dataset collection provides comprehensive daily trading data designed to evaluate quantitative stock selection, portfolio optimization, and algorithmic trading models. It was specifically curated to support research on risk-aware financial decision-making and deep learning-based market forecasting, serving as the benchmark for evaluating the Mean-Variance Neural Utility Function (MV-NUF). The collection comprises four distinct real-world market datasets sourced from the Yahoo! Finance database, encompassing equity markets in the US, Japan, and China:NASDAQ: 1,026 stocks from the highly volatile S&P 500 and NASDAQ indices (Jan 2013 - Dec 2017).NYSE: 1,737 stocks listed on the New York Stock Exchange, representing relatively stable, large-cap assets (Jan 2013 - Dec 2017).TSE: 95 large-cap stocks from the Japanese TOPIX-100 Index (Nov 2015 - Aug 2020).AHK: 85 actively traded stocks across the Shanghai, Shenzhen, and Hong Kong markets (Dec 2014 - Jul 2020).Data Features and Preprocessing: For each asset, the dataset includes five daily trading features: Opening price, High price, Low price, Closing price (OHLC), and Trading Volume. To facilitate immediate use in machine learning models, the data has been preprocessed using z-score normalization. Furthermore, each dataset is strictly partitioned chronologically into training, validation, and test sets to simulate realistic trading scenarios and prevent look-ahead data leakage.This standardized collection serves as a robust benchmark for researchers developing temporal graph neural networks, sequential forecasters, and models integrating Modern Portfolio Theory (MPT) with deep representation learning."