Traditional optimization techniques using mean-variance analysis may result in unstable weights and poor out-of-sample performance. Risk-based methods such as risk parity or equal risk contribution provide more stable portfolios, but they do not consider expected returns.
Our approach combines advances in computing power, the widespread availability of data, and recent academic and internal research to re-incorporate expected return estimates in portfolio optimization. We seek to provide investors with better performance compared to mean-variance or risk-based strategies.
We combine insights from cutting-edge statistical learning methods and decades of market experience.
This scientific approach to dynamic asset allocation leverages our sophisticated computer algorithms and aims to make the best of macroeconomic forecasting and factor investing.