Synthetic benchmark dataset for a FinBERT news sentiment trading pipeline (midcap equities)

This dataset supports the paper "A FinBERT-Based Pipeline for News Sentiment Trading Signals in Midcap Equities: A Proof-of-Concept with Frictions-Aware Backtesting." It contains a synthetic firm-day panel (120 firms over 520 trading days, 62,280 rows) generated with a fixed random seed (seed 11), together with the analysis notebook that produces all tables and figures. The data are synthetic and reproduce the structure of the intended dataset; they are used to validate that the analysis pipeline is correctly implemented and are not real market data.