- Conducted a comprehensive time‑series analysis on 800+ weekly global gold price records (Secondary data) using ARIMA and identified ARIMA (0,1,0) as the most statistically defensible model through ACF/PACF, BIC, RMSE, MAE, and Ljung–Box diagnostics.
- Performed full modelling workflow including auto correlation, stationarity testing, differencing, model comparison, parameter estimation, residual diagnostics and short‑term forecasting.
- Built forecasting outputs and visualizations to interpret price trends and model behavior which demonstrated strong model fit and practical relevance for investment decision‑making.
- Compared the forecasted values of Gold (XAUUSD) and Silver (XAGUSD) using the best ARIMA model to determine which precious metal offers a better RRR in the short term (26 weeks) with a 95% confidence interval.
- Authored a full academic research report for the final year BS project and documented the complete methodology, analysis, results, forecasting outputs and investment implications which demonstrated strong analytical reasoning and professional reporting skills.
Final Year BS Project – Evaluating ARIMA Models for Gold Price Forecasting
A Comprehensive Time Series Study (2010 – 2025)
March 6, 2026
by
Shaikh Ahmed
in Python
Shaikh Ahmed
March 6, 2026