Foreknowledge of earnings surprises is more valuable than trading on known earnings surprises. Earnings surprises are predictable to a considerable extent. I evaluate the performance of an earnings surprise predictor (ESP), which foretells how close analyst expectations of quarterly earnings numbers will be to upcoming earnings numbers. I examine performance over nine years, adjust returns for those of the S&P 500 index, and I adopt an implementable trading strategy. I show that ESP outperforms the S&P 500 index in all nine years. I also show that a weighted portfolio assigning higher weights to stocks expected to have the largest positive earnings surprises has a higher return and a lower variance than a portfolio that assigns equal weights to all rank groups.