Banks need enough cash in ATMs and branches to serve customers, but excess cash creates cost and idle liquidity while shortages damage customer experience. Today, cash planning often depends on static limits, historical averages, and manual judgement rather than demand-aware forecasting.

Banks have historical ATM, branch, payroll, calendar, and transaction data. Time-series forecasting and anomaly detection can turn this data into daily operational decisions for treasury, cashiers, ATM operations, and cash-logistics teams.

A forecasting and optimization layer for ATM and branch cash liquidity. The product should combine historical cash-in/cash-out data, transaction patterns, customer-flow signals, payroll days, holidays, seasonality, location data, and real-time status monitoring to recommend cash limits, replenishment schedules, cash-desk allocation, and collection routes.

• Start with cash-demand forecasting for a small set of ATMs or branch cash desks.
• Real-time ATM status monitoring: online/offline, cash level, transaction activity, technical failures, and alerts.
• Build a dashboard showing current cash balance, predicted demand, shortage risk, excess-cash risk, and recommended action.
• Optimize replenishment and collection schedules for cash logistics teams.
• Add ATM profitability and location intelligence: where to place, move, or resize ATMs based on usage and cash demand.
• Use anomaly detection to flag unusual withdrawal patterns or sudden demand changes.
• Connect payroll calendars, holidays, month-day patterns, and seasonal events as forecasting features.
• Forecast branch cash-desk demand using customer flow, payroll cycles, holidays, day-of-month patterns, and seasonality.
• Recommend cash limits that reduce both excess idle cash and cash shortages.
• Support treasury and cash-logistics teams with replenishment and collection planning.

• Data analytics or location intelligence team
• Experience with geospatial data, BI, forecasting, or operational optimization
• Ability to connect business metrics to location decisions
• Good dashboard and data visualization skills

• Historical transaction and balance data ingestion or synthetic-data upload
• Forecast by location: ATM, branch, or cash desk
• Cash shortage and excess-cash risk indicators
• Recommended cash limit and replenishment amount
• Cash logistics schedule suggestion
• Dashboard for treasury, cash operations, and branch managers
• Forecast accuracy report and backtesting view
• ATM switch / monitoring integration
• GIS-based ATM location analysis
• Incassation route and refill plan optimization
• Alerts for cash-out, technical failure, and abnormal demand
• Branch cash-desk forecasting screen with expected cash gap and risk alerts
• Treasury view for excess cash, shortage risk, and recommended cash limits
• Cash logistics planning for both ATM replenishment and branch cash desks

• The model forecasts cash demand with useful accuracy on backtesting data.
• Operations users can see where a shortage or excess balance is likely before it happens.
• The system recommends a concrete action: replenish, collect, adjust limit, or monitor.
• The prototype shows how idle cash and emergency replenishment can be reduced.
• ATM placement or cash-limit decisions become data-driven instead of purely manual.

• Forecasting must be explainable enough for operations teams to trust it.
• The system should not optimize only for zero shortages; it must also reduce excess idle cash.
• Real-time monitoring and auditability are important for production use.