What We Do

   At BEST, our mission is to research, develop and apply the best Bayesian investment technology on behalf of institutional investors.

The Approach

   We use a Bayesian approach to portfolio management and control. The eponym ‘Bayesian’ refers to the 16th century mathematician Thomas Bayes (click here for a brief overview of Thomas Bayes) and is simply the art and science of making the best decisions by combining new and old information using conditional probabilities. A distinct characteristic of our methodology is that the predictive volatilities and correlations across instrument performance is based on forecasting accuracy, as opposed to historical deviations from the average, and it places risk management at the center of our asset allocation process. The result is a forward looking dynamic process that includes error learning, and does not define risk in the traditional way.


Dynamic Modeling

   Dynamic modeling allows our portfolio to adapt to change as it occurs, either slowly or suddenly. Sudden structural changes typically lead to large forecasting errors. These large errors act as a red flag to the model, telling it that something has changed, that the uncertainty has increased, and that it needs to learn the new underlying relationships. However, if the forecast errors persists, the model adjusts to those relationships in its attempt to minimize future forecast errors. This dynamic modeling is implemented via our proprietary custom-built software.