Harry Markowitz who shared the Nobel memorial prize in 1990 is widely regarded as the founder of the ‘don’t put all your eggs in one basket’ portfolio allocation theory. He showed how investors could pick an optimal portfolio of assets, minimising risk for any given expected return, or maximising expected return for any given risk.
One alternative to following the Markowitz method of portfolio allocation is a simple equally weighted portfolio, known as a 1/ N strategy, for example 50% in stocks and 50% in bonds. Tim Harford, in a recent article compared the complex Markowitz and simple 1 / N methods. One major drawback of the Markowitz theory is that you must know the distribution of returns for all assets in which you are investing. With this knowledge it is possible to calculate an efficient frontier, but without it, you can only guess the frontier. Estimating the distribution of returns has problems, such as the fact that historical correlations are poor predictors of future correlations, and the difficulty of estimating the probability of rare events. 2009 research by DeMiguel, Garlappi and Uppal showed that unless a huge amount of historical data is available (500 years for a 50-asset portfolio), the simple 1 / N strategy outperforms the far more complex Markowitz based calculations implying that the simplest diversification is often the way to achieve the best returns.