We propose an estimate of the optimal portfolio weights under parameter uncertainty. In theory, we prove that our new estimate is close to the true value with an upper bound. Empirical evidence shows that our new estimate can consistently outperform the naive diversification method (1/N Rule),...