This record contains the preprint version of “Stock Pattern Assistant: An Explainable AI Framework for Historical Stock Pattern Extraction and Event Correlation Using Public Market Data.” The paper introduces a fully deterministic and explainable workflow for analyzing historical stock price behavior using only publicly available OHLCV data. The framework identifies monotonic upward and downward price runs, aligns these segments with temporally proximate public market events, and produces concise natural-language summaries using guardrail-constrained large language models (LLMs). The system is designed to maximize interpretability, reproducibility, and regulatory safety. It avoids predictions, forward-looking statements, and causal claims, focusing strictly on historical behavior. Evaluations across four equities—AAPL, NVDA, SCHW, and PGR—demonstrate consistent directional insights, structured event context, and clear explainability across varying volatility profiles. This manuscript is the author’s original preprint, posted prior to peer review and independent of any final publication or conference proceedings. It does not include IEEE formatting, and the final published version may differ.