Barclays has integrated Ant International’s Time-Series Transformer (TST) AI FX Model into its FX hedging operations to help businesses reduce foreign exchange risk and optimize global treasury management.
Built on a transformer architecture and trained with over 2 billion parameters, Ant’s proprietary TST Model leverages advanced time-series forecasting algorithms to deliver hourly, daily, and weekly FX and cash flow predictions with over 90% accuracy. The system continuously improves through pre-training and supervised fine-tuning, enabling more accurate decision-making amid market volatility.
Barclays has already completed the first set of intra-group FX transactions using the AI model, now fully embedded in its BARX NetFX platform, which serves e-commerce and payments clients globally. The move allows Barclays to lower hedging costs while enabling Ant to pass those savings on to its business customers via more competitive rates.
“Their state-of-the-art AI model has improved the accuracy of forecasting cash flows and helped us optimise the FX hedging process,” said Ben Parkinson, Head of Global Fintech & FX Automation Sales at Barclays.
The current implementation covers major currency pairs, with plans to expand to a wider range of currencies and business cases as the partnership grows.
“This collaboration is a strong testament to how Barclays is dedicated to evolving alongside our partners,” added Pushkaraj Gumaste, Head of Corporate Banking for Asia Pacific & Middle East at Barclays.