Technical analysis, also known as "charting," has been a part of financial practicefor many decades, but this discipline has not received the same level of academicscrutiny and acceptance as more traditional approaches such as fundamental analysis.One of the main obstacles is the highly subjective nature of technical analysis-the presence of geometric shapes in historical price charts is often in the eyesof the beholder. In this paper, we propose a systematic and automatic approach totechnical pattern recognition using nonparametric kernel regression, and we applythis method to a large number of U.S. stocks from 1962 to 1996 to evaluate theeffectiveness of technical analysis. By comparing the unconditional empirical distributionof daily stock returns to the conditional distribution-conditioned on specifictechnical indicators such as head-and-shoulders or double-bottoms-we findthat over the 31-year sample period, several technical indicators do provide incrementalinformation and may have some practical value.