Authors - Hetansh Shah, Hitarth Bhatt, Jay Topiwala, Hitanshu Shah, Pradnya Saval, Shruti Mathur Abstract - This paper compares classical and deep models for forecasting stock prices of India’s top energy stocks (ONGC, NTPC, RELI) under inflation stress measured by the Energy Price Index (EPI). Although classical models like Prophet and Holt-Winters accurately forecast the inflation series, an LSTM network with an Attention mechanism forecasts much better for stock price forecasting, especially for volatile stocks. The paper further demonstrates that pure-energy stocks (ONGC, NTPC) are more inflation-sensitive than diversified RELI. The findings have pragmatic implications for investors and policymakers on risk management in the energy market.
