Since the pathfinding research by Markowitz in 1952, various optimal asset-allocation models have emerged. However, those elegant but complicated models have shown poor performance relative to a simple “1/N” strategy due to estimation errors in the U.S. Market. In this thesis, comparing is made that three measurements of 12 asset-allocation strategies to an “1/N” portfolio using historical data generated by the Korean stock market from January 2000 to December 2015. It is found that in the Korean stock market, minimum-variance portfolios with short sale constraints, mean-variance portfolios with short sale constraints, and Bayes-Stein portfolios with short sale constraints perform better than the other models in terms of Sharpe ratio and Certainty-equivalent returns. While these optimal asset-allocation models have higher turnover than the “1/N” portfolio, their Sharpe ratios are still higher than the “1/N” portfolio even after I consider transaction cost. I simulate 2000 years of excess returns data to examine the relationship between estimation windows and Sharpe ratios. Simulated data results show that the Sharpe ratio of the mean-variance portfolio becomes higher when estimation windows become longer. However, the Sharpe ratio of the mean-variance portfolio with short sale constraints and the Sharpe ratio of the Bayes-Stein portfolio with short sale constraints become worse when the estimation windows are extended further.