Balance Forecasting - Financial Institution
Project Information:
I (Mohamed Naji - Owner) built a full balance forecasting and optimization system for a major banking institution as part of a secured, NDA-protected engagement. The project focused on improving liquidity planning, forecasting accuracy, and internal reporting efficiency across multiple financial products.
To deliver this, I developed an end-to-end Python forecasting pipeline and paired it with optimization models to support treasury-level decision-making.
Key contributions included:
- Developed a time-series forecasting engine using Python (Pandas, NumPy, scikit-learn, Statsmodels) to produce seasonality-aware and trend-adjusted balance projections.
- Built a Quadratic Programming (QP) optimization module that recommended balance allocations under regulatory, liquidity, and risk constraints.
- Created, automated, and validated workflows using Excel to cross-check model outputs, compare forecast details, and perform sensitivity testing.
- Implemented automated data pipelines that retrained models based on updated financial data to ensure accuracy over time.
- Designed Tableau dashboards for leadership to monitor forecasts, product-level performance, scenario analysis, and variance tracking.
- Ensured full compliance with strict NDA, data-security, and confidentiality requirements throughout development.
- Improved end-user workflows by reducing manual reporting time, enhancing model consistency, and increasing trust in forecasting outputs.
This project ultimately provided the client with a more reliable forecasting foundation and a scalable framework for future growth in analytics and treasury operations.
