Technical trading rules use price and volume data to forecast future returns. The authors test the ability of widely available technical indicators to forecast the daily price returns of the DJIA. They apply quantitative methods to identify apparently successful technical trading rules, compensate for false positive indications of trading rule value, examine out-of-sample performance persistence, and make temporal adjustments to transaction costs. They include extensive references to previous academic research on technical trading strategies as well as appendices of formulas, methods, and assumptions.

Technical indicators are widely available and heavily promoted by popular finance websites, online retail brokers, and market data providers. Therefore, a dispassionate investigation of their theoretical and real-world efficacy is valuable to analysts, strategists, traders, portfolio managers, and risk managers. The authors’ novel application of the false discovery rate (FDR) incorporates the possibility that technical indicators that have no genuine predictive value will sometimes appear profitable. They also develop a methodology for decision making under the condition of multiple conflicting signals. Finally, in-sample and out-of-sample performance tests for persistence and a rigorous examination of transaction costs result in an impartial and thorough analysis of technical trading strategy performance.

The authors conclude that publicly available technical indicators currently do not have predictive power that exceeds their transaction costs. But they do concede that nonpublic technical indicators may exist that have predictive value and that historically there have been periods when technical indicators appear to have predictive value before transaction costs. Although some rules do show the potential to outperform, these rules are difficult to select ex ante.

Using daily prices of the DJIA from January 1897 to July 2011, the authors reexamine 7,846 previously researched technical trading rules. These rules are from several categories, including filter rules, moving averages, support and resistance, channel breakouts, and on-balance volume. They implement a novel application of the FDR methodology to account for data snooping and to identify outperforming technical indicators. The FDR methodology uses technical trading rules that truly add value and discards rules that appear to have positive performance but have been selected by chance or by error. They are then able to quantify the possibility that some of the selected rules are the consequence of false positive test results and do not have any genuine predictive value. The subsequent tests use only the rules that have survived the FDR filter.

The authors further address deficiencies in previous research on technical trading rules through a detailed examination of transaction costs and their changing impact over time as a result of financial innovation. The addition of transaction costs renders most short-term trading strategies unprofitable. Rules deployed before 1962 may have been profitable if transaction costs were below 16 to 70 bps. Since 1962, most strategies have been unprofitable, even with the assumption of zero transaction costs.

Although technical indicators receive a great deal of publicity, the authors find they have no economic value after accounting for transaction costs. The method they use is a novel adaptation of a highly quantitative method previously used to examine the value and persistence of mutual fund selection. My primary concern centers on the exclusive focus on the DJIA. Additional tests on other equity, fixed-income, and commodity time series as well as coverage of other countries would resolve the question of technical indicator value with greater confidence.