Predicting foreign exchange rates is a complex science but paramount to investors. An investment banking client required accurate, rapid Forex forecasting, to optimise and balance expectancy and risk spread in client portfolios, to maximise investor returns.
Transition from Excel based system to integrated web-based client
Replace traditional back and forward testing approach with ML based predictions
Maintain security of client data and proprietary algorithms
Optimising for rapid processing despite huge computational complexity
Deliver very high accuracy in predictions
1. Data Architecture Analysis
Bigquery database for secure client data caching
Secure front end web-client
SQL, python and R-based ML pipeline
Realtime data processing and scrapping
2. Technical Implementation
Machine learning algorithm optimisation
Validation with 14 years of past data
Beta testing and iterative design with end users
Faster insight discovery and vast reduction in manual data entry
Improvement in recovery of audit trail
Significantly improved accuracy of predictions
Attracting additional clients
Forward opportunities opened for forecasting in other investment spaces
System ready for new market expansion
"The new software is a game changer for what we do. We can now service more customers and report higher accuracy to our investors"
Investment Banking Client
Initial Development: 12 months
Validation and beta testing: 6 months
Ongoing optimisation and development
Expanding into sports forecasting and heavy metals models.