In today’s rapidly transforming global economy, the levels of risk and return of different assets are highly time-variant. Finding an optimal asset allocation strategy is a complex task which needs to be continuously adjusted.
This white paper explores how machine learning, in particular deep learning and evolutionary algorithms, can be employed to create a global and flexible asset allocation model which can adapt to different market conditions over time. It is based on all asset classes and regions, and takes into account prices and risks, correlations, and market sentiment. The paper also describes how the underlying deep neural networks can be trained with maximum performance using CPUs and GPUs.
- Flexible multi-asset allocation model
- Using deep learning and evolutionary algorithms
- Maximising absolute return performance
- Adapts to changing market parameters
- Fast training on CPUs and GPUs