Insider Brief
- Terra Quantum and BBVA completed a successful pilot project demonstrating the use of AI-driven and quantum-inspired methods to enhance exotic derivatives pricing efficiency.
- The project achieved millisecond-level pricing with high accuracy and scalability, utilizing advanced computational methods like Neural Networks and Tensor Network Layers.
- Significant efficiency gains were realized, including a 260x faster inference on standard CPU hardware through model compression techniques.
PRESS RELEASE — Terra Quantum and BBVA have successfully completed a pilot project, demonstrating the potential of AI-driven and quantum-inspired methods to enhance the speed and efficiency of exotic derivatives pricing.
Structured products, such as exotic derivatives, involve complex payout structures and require accurate, high-speed pricing across a wide range of market conditions and parameters. Advanced computational methods are essential to efficiently navigate these high-dimensional problems while meeting the accuracy and performance demands of financial institutions.
The pilot explored innovative approaches, including Neural Networks and Tensor Network Layers, to address the challenges of rapidly pricing these complex products, considering broad parameter ranges and granular increments for exotic derivatives covering several underlying assets. The results show that these techniques can achieve significant improvements in computational efficiency and speed, while maintaining accuracy.
Key Achievements
Through the collaboration, Terra Quantum demonstrated:
- Improved Speed and Accuracy: Achieved millisecond-level pricing while meeting target accuracy thresholds across various test scenarios.
- Scalability: Successfully managed large parameter spaces, aligning with the requirements of complex exotic derivatives.
- Efficiency Gains: Model compression techniques enabled 260x faster inference on standard CPU hardware, improving computational efficiency and cost-effectiveness
While alternative approaches, such as Tensor Train Cross (TT-Cross), provided valuable insights, the project highlighted the greater robustness and flexibility of Neural Networks for handling real-world complexities at scale.
A Step Towards Future Applications
The successful completion of this pilot highlights the potential for advanced AI and quantum-inspired methods to streamline pricing tools for financial institutions. Terra Quantum’s solution demonstrates clear opportunities for further development and deployment in production environments.
“We are pleased to be collaborating with an industry leader in quantitative methods, such as BBVA,” said Vishal Shete, Managing Director at Terra Quantum. “The results confirm that AI-driven approaches, combined with quantum-inspired techniques, can deliver meaningful improvements in complex pricing problems, with clear pathways for further optimization.”
“As a bank heavily focused on technological innovation, at BBVA we need to test out the latest cutting-edge technologies to help optimise complex financial calculations,” explains Escolástico Sánchez, Head of Quantum at BBVA. “The pilot we have just completed with Terra Quantum is a further step in this strategy.”
