We ask what determines the profitability of candlestick trading strategies. Is it the definition of trend and/or the holding strategy that one uses in candlestick charting analysis? To answer this, we systematically consider three definitions of trend and four holding strategies. Applying candlestick trading strategies to the DJIA component data, we find that regardless of which definition of trend is used, eight three-day reversal patterns with a Caginalp–Laurent holding strategy are profitable when we set the transaction cost at 0.5% and after we account for data-snooping bias, while the patterns with a Marshall–Young–Rose holding strategy are not profitable. For sensitivity analysis, we also find that our results are not qualitatively changed on a lower transaction cost of 0.1%, or when we conduct the subsample analyses based on three equal periods and three distinct market conditions. When considering a more volatile market, evidence in favor of candlestick trading strategies is strengthened.

We propose a Cramér–von Mises–type test for testing whether the mean potential outcome given a specific treatment level has a weakly monotonic relationship with the continuous treatment under unconfoundedness. To flexibly control for a possibly high-dimensional set of covariates, our test is based on a double debiased machine learning method. We show that our test controls asymptotic size and is consistent against any fixed alternative. We apply our test to evaluate the Job Corps program and reject a weakly negative relationship between the treatment (hours in academic and vocational training) and labor market performance among relatively low treatment values.