Japanese Candlesticks have been in use for centuries. Though detractors of the discipline of technical analysis liken it to entrail reading, many a trader swears by candlesticks as powerful predictors of market direction – many even claim them profitable. This piece discusses two separate studies conducted on classic markets, that scrutinized the predictive efficacy of Candlestick patterns.
Munehisa Honma, an 18th century Japanese rice trader, is credited as the developer – or perhaps the discoverer – of either, Japanese Candlesticks, or the methods that modern traders use so widely today, accounts tend to vary. His discovery, likely a rougher version of Elliott’s Wave Principle, was that Candlestick Patterns offered insight into future market action. His brilliance was in developing a strategy to use the 18th century Japanese rice market’s Composite Man to extract massive gains from his trading activity.
Most traders play candlestick patterns, largely, by ear and rely on a combination of knowledge and intuition to derive profits from their trading actions. However, members of the world of Mathematics, and Science have, over the years, employed various quantitative methods to come to a final conclusion as to whether Candlestick Patterns really do offer any of the predictive power they are famed for. The results have been mixed.
We look at a research piece by Piyapas Tharavanij, Vasan Siraprapasiri, and Kittichai Rajchamaha, which examines the Profitability of Candlestick Charting Patterns in the Stock Exchange of Thailand by borrowing a good amount of insight from studies conducted over the previous two decades, or so. The other piece studies the predictive efficacy of Japanese Candlestick Patterns in relation to Forex markets. Authored by Ismael Orquín-Serrano, the article looks into the Predictive Power of Adaptive Candlestick Patterns in Forex Markets, in the case of a EUR/USD market pair.
The Research
Study One:
The research piece by Tharavanij et al seeks to determine the profitability of Candlestick Patterns, by backboning its analytical processs on previous works that cast an eye at the notoriously mysterious price/time ratio indicators. Notable references to research by Caginalp and Laurent (1998), Marshall, Young, and Cahan (2008), who examined different Western Equity Markets in relation to the question, and Zhu, and team (2016), who applied their study of Japanese Candlestick reliability to Chinese markets. This particular team found that the bearish harami pattern showed itself a reliable reversal indicator.
Research of this nature is inspired by a 1980 challenge – the first of many – to the Efficient Market Hypothesis (EMH), which posits that the price of an asset already has all the relevant information factored into it and therefore, chart patterns, as well as other analytical methodologies relying on historical data to extract profit from the market’s future movements, is much like astrology or tarot card reading.
Tharavanij and cooperators applied a Skewness Adjusted Test, to determine the profitability of candlestick patterns over different holding periods (1,3,5, and 10 trading days) as they identified skewness as having a more significant effect on statistical distribution than Kurtosis would. Their analysis relied on Filtering, which is the use of indicators to confirm candlestick signals.
Study Two:
Possibly the more significant of the two – as Forex markets are often viewed as being the most characteristically similar legacy markets, to cryptocurrency markets – is the research article published by Ismseal Orquin-Serrano, in April of 2020. Orquin-Serrano, not only, sought to test the validity of Japanese Candlestick Signals, but to ascertain whether – or not – Forex markets are EMH efficient. To know – for sure – if Forex markets follow a purely stochastic price development process, or not, the researcher examined the rather liquid, EUR/USD asset pairing, considering prices only.
Orquin-Serrano details categorizing various Japanese Candlestick types by size (Small, Medium, Big). The study took volatility changes into account, in order to confirm that different conditions of volatility can be identified in the Candlestick Patterns.
“If the trading strategy has no predictive power, its average return is zero,” is the premise guiding this research piece. The researcher analyzed a Candlestick based Stop-Loss/ Take-Profit strategy. Even going as far as incorporating machine learning to see if AI can play a sufficient role in aiding a trader’s predictions.
The Results
Tharavanij, and the gang found that there was little predictive usefulness to Candlestick signals, bullish, or bearish. This being due to the fact that the mean returns on examined patterns didn’t deviate too far from zero. They also found that signal strength, or direction (whether bullish, or bearish) weren’t, altogether, textbook either.
They do admit, however, that the study’s limitations may lie in the fact that TA standards, like support, and resistance, or trend, where not taken into account – which leaves the Elliott Wave Principles out. The study concludes by suggesting that Candlestick Patterns may be more effective with smaller capitalized stocks.
Ismael Orquin-Serrano’s analysis turned out similar results. Using his method of analysis, the EMH holds true. With Learning Algorithms being slightly better predictors than humans.
Closing Thoughts
Although both of the studies found that Candlestick Patterns yielded no, statistically significant, predictive edge over future market movements, both also stated that the result they came to could be a result of the parameters employed in their research. Top traders tend to use candlestick patterns, in tandem with other trading disciplines and tools to come to a trading decision, perhaps that is what led to their conclusions.
Honma, it is said, only had the candlesticks to go with – unless a crowded rice market was his volume indicator – so maybe candlestick reading does not belong to the realm of statistics and mathematical absolutes, but requires a degree of intuition – as they are generated by the market actions of living people.
What interests me most is the statement by Tharavanij et al, suggesting that Candlestick patterns may hold more predictive power in smaller capitalized markets. The digital assets market, in its current form, probably more than fits the bill. This is an open call to researchers for more studies – of this nature – based on crypto markets. How do candlestick patterns play out in crypto? What are Bitcoin Market characteristics in relation to the Elliott Wave Principle? Perhaps, polish up some research methodologies that lead to more conclusive answers?
