Since Graham and Dodd (1934), top finance professionals have argued that value investing— the orderly pursuit of underpriced securities—delivers a premium by outperforming the market over time. Today, value strategies are popular with equity investors. However, there is no single way to construct a value strategy, and the magnitude of the value premium depends on the value metric used, the implementation, and the investment period. In this study, we investigate how the choice of accounting metric and implementation affect the performance of a value strategy. Our study generated several empirical insights. Strategies based on book-to- price (B/P) and earnings-to-price (E/P) ratios delivered a positive premium over the 60-year horizon from 1951 to 2013, with E/P having higher return and lower risk than B/P over the full horizon. However, B/P outperformed E/P between 1963 and 1990, and that was the basis of the landmark study establishing B/P as the academic standard. Strategies based on a blend of B/P and E/P outperformed both single-metric strategies during most 10-year periods between 1973 and 2013. Over the same horizon, optimized value strategies had lower tracking error, lower turnover, and a higher information ratio than “rank-and-chop” strategies, which weight high-percentile value stocks by capitalization. Sector constraints raised both the Sharpe ratio and the information ratio of an optimized blended-value strategy.