The growing complexity and volatility of financial markets have increased the need for practical, data-driven, experiential learning in business and finance education. Yet conventional instruction often relies on static cases or historical data, limiting exposure to real-time market dynamics and behavioral decision-making. This paper presents FaiseStock, a real-time paper trading platform for university education that uses live Vietnam market data and is publicly accessible at: https://stock.faise.vn . FaiseStock supports realistic trading, including market orders, order matching, portfolio and asset management, contest-based learning, and learning analytics. The platform is built on a modular, event-driven architecture with an integrated pipeline that converts trading events into structured behavioral datasets. Using 1,501 orders from 107 active users across 98 stock symbols, we analyze trading behavior and engagement. We observe buy-side dominance: 72.0%, a preference for limit orders: 61.4%, and an overall completion rate of 69.3%. Market orders were fully completed: 100%, while pending/cancelled orders occur mainly among limit orders. Participation heterogeneity, captured by the “orders per user” distribution, provides an engagement signal. Overall, the study contributes (i) a reference architecture and data pipeline for real-time educational trading systems, (ii) behavioral evidence from student interactions, and (iii) implications for experiential and competition-based learning design.