https://orcid.org/0000-0003-2774-5777
![Sample our Economics, Finance,Business & Industry journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5931/TOC-Subject-Sample-2021-Economics-Finance-Business-Industry.jpg"})
Abstract
This paper explores an unexamined sentiment channel through which technical analysis can add value. We use a spectrum of technical trading strategies to build a daily market sentiment indicator that is highly correlated with other commonly used sentiment measures. This technical-analysis-based sentiment indicator positively predicts near-term returns and is inversely related to long-term returns in the cross-section. Simple trading strategies based on this sentiment indicator yield substantial abnormal returns. These results are consistent with the explanation that lack of synchronization induces rational arbitrageurs to exploit the mispricing before it is corrected.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Supplemental data
Supplemental data for this article can be accessed online at https://doi.org/10.1080/14697688.2023.2244991.
Notes
1 For example, Sushil Wadhwani, an academic who later became a fund manager, once said that overcoming the prejudice against technical analysis was the most important lesson she had to learn when moving from the ivory tower into the laboratory of real-life experience as a trader. See ‘Technical analysis pulled out of the bin’, October 17, 2010, Financial Times. It is also more generally prominent in the financial media, e.g., the 200-day moving average was cited four times in the widely-read daily ‘Market Forces’ bulletin of the Financial Times during December 2019.
4 Investor sentiment indicators are abundant and well accepted at the market level, whereas they are scarce at the individual stock level. To test the correlation of TA sentiment and other sentiment indicators, we restrict ourselves to market-wide TA sentiment in this paper.
5 Since it is often difficult to predict the exact time when delayed and coordinated arbitrage occurs, our trading strategy is designed to exploit the sentiment-raised momentum rather than the profit arising from the correction of mispricing that occurs when the triggered arbitrage brings stock prices to their fundamentals.
6 The definitions of these trading strategies are the same as in Qi and Wu (Citation2006), and are standard in the literature. Therefore, they do not merit further elaboration here. The parameters used for defining those technical trading strategies are provided in the Appendix.
7 We also calculate a performance-weighted TA sentiment index, which is the average of the trading signals of 2127 technical trading rules weighted by their returns in the past year. Our results do not alter when using the performance-weighted TA sentiment index.
8 The TA sentiment index is available upon request.
9 The Baker and Wurgler sentiment index is positive for 1968–1970, 1972, 1979–1987, 1994, 1996–1997, and 1999–2001.
10 Most of the other randomly selected samples exhibit similar co-movement.
11 VIX measures the market expectations of the volatility conveyed by S&P 500 stock index option prices over the next 30-day period. Put-Call ratio is a ratio of put volume to call volume and is a contrarian indicator of market sentiment. Individual Bull/Bear spread is the percentage of individual investors who are bullish minus the percentage of individual investors who are bearish about the stock market for the next six months. Individual Bull/Bear spread based on data from the American Association of Individual Investors, which polls opinions of AAII members on weekly basis, and is available from http://www.aaii.com/sentimentsurvey.
12 The data are available at http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html. RMRF is the market return premium over the risk-free rate, SMB is the average return on the three small portfolios minus the average return on the three big portfolios, HML is the average return on the two value portfolios minus the average return on the two growth portfolios, RMW is the average return on the two robust operating profitability portfolios minus the average return on the two weak operating profitability portfolios, and CMA is the average return on the two conservative investment portfolios minus the average return on the two aggressive investment portfolios. The momentum factor (UMD) is the average return of high prior return portfolio over low prior return portfolio.
13 We set 10 as the maximum lag to be considered in the autocorrelation structure when calculating Newey-West robust standard errors for the coefficients.
14 These are the only microeconomic variables for which data is available on a daily frequency. Both the Default spread and TED spread are from https://fred.stlouisfed.org/. Default spread is the difference between Moody's AAA and Baa bond yields and TED spread is the spread between the three-month LIBOR based on US dollars and the three-month Treasury Bill.
15 The data is downloaded from https://www.philadelphiafed.org/research-and-data/real-time-center/business-conditions-index.
16 The daily EPU data is available from https://www.policyuncertainty.com/us_monthly.html.
17 In unreported results, we regress the one-month leading S&P 500 index returns (including dividends) on Neely et al.’s (Citation2014) F1_TECH, our monthly aggregated TA indicator, and PCA_TA indicator, respectively. We find that both the monthly average of TA indicator and the aggregated PCA_TA indicator are poor predictors of the aggregate market return. However, based on the significance of the parameters as well as the adjusted R-squared, F1_TECH is a better predictor of the monthly aggregate monthly return than TA. This is consistent with our earlier argument that the TA indicator captures better the short-term investor sentiment and hence is more profitable in the cross-section rather than in the aggregate market. Our arguments and evidence are consistent with the argument in Baker and Wurgler (Citation2006) that sentiment indicators are poor predictors of the aggregate market return. Further details on these results can be obtained from the authors.
18 We extend the tests of the predictive power of TA beyond the conditional market beta model. If TA sentiment is a state variable that captures the changing conditions in macroeconomics and financial market, this can be reflected in the conditional factor loadings that are linear in the lagged TA sentiment. This would allow factor loadings to vary over time and correlate with the TA indicator. For this purpose, we estimate a conditional four-factor Carhart model. We show that while our TA sentiment indicator influences the beta loadings of Carhart four factors, it does not alter signs or the significance of the coefficients on the lagged TA indicator. Thus, the predictive power of our TA sentiment indicator could not be fully subsumed by the changing beta loadings of other pricing factors. Details are available upon request.
19 It has been pointed out in Baker and Wurgler (Citation2006) that ME portfolio return premium strongly correlate with other portfolio returns. When the dependent variables are equally weighted, the size effect might play a large role.
20 We consider moving average of alternative numbers of days (1, 5, 30, 60, 120, 250) in our robustness check.
21 We can also apply our timing strategy to individual decile portfolios as in Han et al. (Citation2013). Returns (BETC) on the TA timing strategy are much higher (higher) for the most sentiment-prone deciles than that of the long-short portfolio constructed with the same firm characteristic.
22 We have also repeated the analysis using monthly data. We find that the aggregated monthly TA sentiment generally has a weaker relationship with the next-month return, suggesting that TA indicator predicts different patterns of returns across the short- and long-term. Consistent with prior literature on investor sentiment as a contrarian predictor of cross-sectional return, we also find that our TA sentiment indicator is negatively related to the next three-month returns. We also calculate the performance of a trading strategy based on monthly TA sentiment. We find that the benefit of using TA sentiment on monthly frequency disappears, showing that the sentiment-driven momentum fades away in less than 10 days. The details of these results are available upon request.