UOB Asset Management improves trading strategies with Vertex AI
UOB Asset Management developed an AI hedging model using Google Cloud Vertex AI to predict financial market movements and enhance investment returns. They refined the model by reducing input features and incorporating economic indicators to improve efficiency and explainability. The improved model reduced trade processing time from 48 hours to 2 hours and outperformed buy-and-hold strategy by 28.7%. The AI enhancements have the potential to generate significant additional returns for investors and provide a scalable framework for future improvements.
- Organization
- UOB Asset Management
- Industry
- Finance
- Location
- Singapore
Reported outcomes
28.7%
quantified impactOther quantified impact
Strategic outcomes
Primary read
Use case focus
Showing 3 of 3
- 1AI for Investment Management
- 2Trading Strategy Optimization
- 3Predictive Analytics
- UOB Asset Management needed to leverage AI to deliver better investment outcomes and reliable market predictions.
- The team required transparent development and integration of AI into existing financial workflows.
- They targeted performance improvements in hedging strategies and faster trade processing.
- Utilized Google Cloud Vertex AI platform to build and iteratively improve a hedging AI model.
- Incorporated 1,000 relevant financial and economic features for model analysis, reducing from 2,000 initially.
- Defined clear framework boundaries and objectives to separate AI-driven performance enhancement from other tools.
- Trade processing time significantly reduced from 48 to 2 hours, boosting operational efficiency.
- Hedging model showed a 4.77% recall improvement and 1.86% F1 score increase.
- AI model outperformed traditional buy-and-hold strategy by 28.7%, potentially delivering substantial investor returns.
- The platform supports ongoing refinement and scaling of AI models within compliance requirements.
Sources & evidence1
AI-generated summary. Verify important details with the linked sources before relying on this case.
Explore related AI use cases
Was this useful?