Scientific Beta

This paper introduces the Scientific Beta methodology for constructing an implementable proxy for the Minimum Volatility portfolio – the least risky portfolio on the efficient frontier. The methodology draws on a number of advances in the literature on Minimum Volatility portfolio construction including i) the reduction in estimation errors by using a suitably designed statistical factor model, ii) the use of flexible constraints on overall portfolio concentration (“norm constraints”), and iii) the use of liquidity and turnover constraints to ensure that the strategy is easily implementable.

This paper introduces the Scientific Beta methodology for constructing an implementable proxy for the Minimum Volatility portfolio – the least risky portfolio on the efficient frontier. The methodology draws on a number of advances in the literature on Minimum Volatility portfolio construction including i) the reduction in estimation errors by using a suitably designed statistical factor model, ii) the use of flexible constraints on overall portfolio concentration (“norm constraints”), and iii) the use of liquidity and turnover constraints to ensure that the strategy is easily implementable. The strategy typically results in a low beta portfolio and hence provides outperformance in bear markets. Using the Fama French three-factor model, we find that the strategy shows significant small cap and style tilts. However, these factor tilts can be controlled by using an appropriate stock selection in combination with the Efficient Minimum Volatility weighting.