Published Online:July 2025
Product Name:The IUP Journal of Accounting Research & Audit Practices
Product Type:Article
Product Code:IJARAP040725
DOI:10.71329/IUPJARAP/2025.24.3.74-85
Author Name:S Ravi Kumar, D I P Mani Kumar and B D Hansraj
Availability:YES
Subject/Domain:Finance
Download Format:PDF
Pages:74-85
The paper evaluates the Prophet model’s effectiveness in forecasting Nifty 50 index trends using 10 years’ daily data (2013-2023) from the National Stock Exchange (NSE). Prophet is selected for its ability to capture nonlinear trends, seasonality, and event-driven fluctuations. The analysis focuses on a short-term 30-day forecast horizon, offering relevant insights for active traders, while highlighting the model’s potential for long-term applications. Forecast accuracy is assessed using mean absolute error (MAE), root mean squared error (RMSE) and mean absolute percentage error (MAPE), with results showing a consistent upward trend alongside volatility during key events such as RBI monetary policy announcements, elections, earnings releases, and global developments. The model also accounts for holiday effects, including Diwali and Christmas, which influence market sentiment and liquidity. While Prophet effectively models seasonal patterns, its accuracy declines during high-volatility periods. The limited forecasting window may restrict its usefulness for long-term investors, suggesting a need for extended horizons and incorporation of macroeconomic variables. The findings offer practical insights for traders and policymakers and provide a foundation for future research comparing Prophet with deep learning or hybrid forecasting models.
Time series forecasting plays a crucial role in financial sector, where predicting future price movements enables traders, investors, and policymakers to make more informed decisions. Stock indices, such as Nifty 50, encapsulate the performance of a selected set of companies and reflect broader economic trends, sector performance and investor sentiment.