Using options to improve resource allocation by 20%

We developed a more flexible, more robust method for valuing complex assets in complex, highly volatile environments that outperforms its benchmark by nearly 20%.

The client wanted to investigate the viability of extending real options methods to risk management settings that are generally not amenable to an options formulation. Armed with a data set of more than 200,000 decisions for the domain under study as well as data on the underlying asset movements, we combined machine learning methods with a simple options model to build an accurate forecaster of asset movements and an algorithm that improves allocation decisions by nearly 20%.

While real options are increasingly well-accepted as instruments to value and invest in R&D projects, their applicability remains limited for the restrictive assumptions they place on uncertainty and for their inability to accommodate many features of complex decisions, like multi-agent competition. We used a very large data set of asset movements in a complex setting to forecast asset prices accurately and to solve the problem of making optimal decisions on the back of these forecasts.

Our model has achieved a number of important milestones. It removes a lot of the restrictions around options modeling, extending the applicability of options to complex assets and environments. It also extends the optimal portfolios generated by the Black-Litterman asset allocation model to asset classes that are not widely traded, like corporate R&D investments. In so doing, like Black-Litterman, our model provides a framework for combining investor views with a global capital market equilibrium. In the aggregate, our model outperforms its benchmark by nearly 20%.

• © QuantumBlack LLP 2009-2011
• Legal Notice

• Twitter
• Facebook
• Linked in
• RSS