Technical analysis seeks to exploit stock mispricings derived from postulated investor/trader psychological biases. Does short-term technical analysis actually produce abnormal returns? Or, do its adherents persist based on a misperception that they are to some degree in control of random rewards. In their February 2006 paper entitled “Does Intraday Technical Analysis in the U.S. Equity Market Have Value?”, Ben Marshall, Rochester Cahan and Jared Cahan investigate whether intraday technical analysis is profitable in the overall U.S. equity market. Specifically, they apply a combination of statistically rigorous bootstrapping tests to 7,846 trading rules from five rule families (Filter, Moving Average, Support and Resistance, Channel Breakouts, and On-Balance Volume). Using 5-minute data for Standard and Poor’s Depository Receipts (SPDR) over the period 1/1/02-12/31/03 (encompassing both bear and bull trends), they conclude that:
- During 2002 (2003), the S&P 500 index loses 21.2% (gains 21.9%), and mean intraday and overnight returns are negative (positive).
- Before accounting for data snooping bias:
- 42 of the 7,846 rules are statistically significant at the 1% level.
- 163 rules are statistically significant at the 5% level.
- The Channel Breakout (On-Balance Volume) rule family is the most (least) profitable.
- All of the rule families have a best rule that is highly statistically significant in 2002, but none are close to being statistically significant in 2003.
- After accounting for data snooping bias, none of the 7,846 rules are profitable, either for the entire two-year period or separately for the 2002 and 2003 subperiods.
- Results do not take into account trading costs, which would make performance of all the rules worse.
The authors provide brief descriptions of the technical analysis rule families.
In summary, there is no evidence of any systematic intraday inefficiencies in SPDR data.
In the closing chapter of his 2007 book Evidence-Based Technical Analysis: Applying the Scientific Method and Statistical Inference to Trading Signals, David Aronson reaches a very similar conclusion about 6,402 technical trading rules tested on the S&P 500 index.
