Instability in minimum variancce frontier

The question goes like this… Analysts attempting to compensate for instability in the minimum variance frontier will find which of the following strategies least effective? 1) gathering more accurate historical data 2)Reducing the freequency of portfolio rebalancing 3)Eliminating short sales Guys, could you just help in cracking this and substantiate… Thanks

gathering more historical data you need better forecasts of data. not better historical data to reduce the instability of the min. variance frontier - the inputs are critical. constraining short sales - makes the weights of the portfolio +ve. reducing frequency of rebalancing - also helps since the cumbersome recalculation, (esp. on big portfolios) and the changes because of that are avoided.