Ant International unveils AI model that predicts trading volumes and cuts hedging costs in financial services

Ant International’s Time-Series Transformer AI model predicts FX trading data with over 90% accuracy, reducing hedging costs and risk premiums. It serves banks, airlines, and e-commerce clients.

Published on: May 28, 2025
Ant International unveils AI model that predicts trading volumes and cuts hedging costs in financial services

Ant International’s Numerical AI Model: The GPT Equivalent for Financial Services

Singapore-based Ant International has developed an AI model specifically for foreign exchange (FX) trading that could have as significant an effect on financial services as GPT has had on business communication and content generation. Their Time-Series Transformer (TST) AI FX Model focuses on predicting numerical data rather than generating text or content.

Kelvin Li, general manager of Ant International’s platform tech business unit, describes this approach as a distinct path in AI alongside large language models (LLMs) like those behind ChatGPT. While many financial firms use LLMs to help reduce risk, these models don’t directly address core financial operations such as trading, pricing, or transaction processing.

Li believes the TST AI FX Model could become foundational not only for financial services but also for digital payments and broader economic activities.

How the TST AI FX Model Works

Ant International, a subsidiary of Hangzhou-based Ant Group (the fintech arm of Alibaba Group), currently uses the TST AI FX Model to forecast its own cash flow and FX exposure. The model generates predictions on an hourly, daily, and weekly basis with over 90% accuracy.

This high level of precision allows Ant International to anticipate trading volumes more accurately, leading to lower hedging costs and reduced risk premiums imposed by banks. The model’s clients include banks, airlines, online travel agencies, and e-commerce platforms.

Upcoming Developments

The company is preparing to announce a partnership with a low-cost airline in the region, signaling further practical applications of the model in industries sensitive to currency fluctuations and transactional risk.

  • Focus on numerical prediction rather than content generation
  • More than 90% accuracy in FX exposure forecasting
  • Reduces hedging and risk-premium costs
  • Supports clients across banking, travel, and e-commerce

For professionals interested in AI applications in finance, exploring specialized AI models like Ant International’s TST could provide a clearer understanding of how AI can optimize trading and risk management processes. For further learning on AI in finance and beyond, consider checking out AI tools designed for financial applications.