Technical AnalysisMarket Efficiency

Do Candlestick Chart Patterns Predict Stock Returns? Evidence from Thailand

Summary by Robert Gorak · Published July 21, 2026 · Last reviewed July 21, 2026

Piyapas Tharavanij, Vasan Siraprapasiri and Kittichai Rajchamaha·2017·SAGE Open
Sample: 50 component stocks of the SET50 indexData: SET50 index component stocks (SETSMART database)Period: July 3, 2006 to June 30, 2016

Candlestick charting is a technical analysis method that plots daily open, high, low, and close prices to reveal patterns believed to signal price reversals. Tharavanij, Siraprapasiri, and Rajchamaha's 2017 paper, "Profitability of Candlestick Charting Patterns in the Stock Exchange of Thailand," tested patterns on 50 Thai SET50 stocks. The 10-year study, covering July 3, 2006, to June 30, 2016, found that most patterns' mean returns were not statistically different from zero. Even the significant patterns, like the Opening White Marubozu's 0.71% return over a 10-day holding period, carried standard deviations up to 8.04%.

What the Study Found

The Opening White Marubozu (OWM) returned 0.71% over 10 days under the MYR exit strategy, with an 8.04% standard deviation. The Bullish Harami (BullH) was the only bullish 2-day pattern significant across all holding horizons, returning 0.84% over 5 days with a 7.04% standard deviation. The bullish Closing White Marubozu (CWM) returned -0.17% over 1 day, a significant negative return despite its bullish classification. Conversely, the bearish Closing Black Marubozu (CBM) returned +0.09% over 1 day, behaving like a bullish signal instead. Filtering with the RSI indicator raised the Long White Candle pattern's profitability from under 1% to over 1% in most cases.

Methodology

The study used daily open, high, low, and close prices for the 50 component stocks of the SET50 index from the SETSMART database. The sample period covered July 3, 2006, to June 30, 2016, a 10-year span. Trends were identified using a 10-day exponential moving average (EMA10), comparing closing price against EMA10 to define uptrends and downtrends. Statistical significance was tested using a skewness-adjusted t test (Johnson, 1978) and a binomial test, under both MYR and CL exit strategies.

Key Statistics

Metric Finding Context
OWM return (10-day, MYR) 0.71% Highest holding-period return among all patterns; SD 8.04%
BullH return (5-day, MYR) 0.84% Only bullish 2-day pattern significant across all horizons; SD 7.04%
CWM return (1-day, MYR) -0.17% Significant negative return despite bullish classification
CBM return (1-day, MYR) +0.09% Significant positive return despite bearish classification
EMA10 trend formula EMA_{N,t} = [2/(N+1)](C_t − EMA_{N,t−1}) + EMA_{N,t−1} Defines uptrend/downtrend used to filter valid candlestick signals
Skewness-adjusted t test t_sa = √n[s + (1/3)γs² + γ/(6n)] Tests whether mean holding-period return differs from zero under skewed returns

Why This Matters

Candlestick patterns appear to offer little dependable edge for traders in Thailand's emerging market. The result extends prior null findings from the U.S. and Japan to a new Southeast Asian market. The few patterns showing statistically significant profits came bundled with risk levels large enough to offset the expected gain in practice. Because signal direction did not always match textbook classification, traders relying on off-the-shelf candlestick rules risk mistaking noise for a genuine trading signal. The findings support market efficiency in the SET50 segment, where historical price patterns alone do not appear to generate reliable abnormal returns.

Frequently Asked Questions

50 component stocks of Thailand's SET50 index were tested for candlestick pattern profitability by Tharavanij, Siraprapasiri, and Rajchamaha (2017), covering 2006 to 2016. Candlestick charting itself is a technical analysis method that plots daily open, high, low, and close prices to reveal patterns believed to signal reversals. Most patterns showed no significant predictive power.

0.71% was the highest statistically significant return found, from the Opening White Marubozu pattern over a 10-day holding period. It carried an 8.04% standard deviation. Most of the candlestick patterns tested showed no statistically significant relationship with future returns in the SET50 index.

Filtering with Stochastics (%D), RSI, or MFI generally did not increase profitability or prediction accuracy for candlestick patterns tested on 50 SET50 stocks. The RSI filter was an exception for the Long White Candle pattern, raising its return from under 1% to over 1% in most cases. Filtering also reduced the number of trades.

Candlestick reversal patterns were tested on Thailand's SET50 index over a 10-year period (July 2006–June 2016). Most produced mean returns not statistically different from zero. The binomial test confirmed that even significant patterns could not reliably predict market direction.

Source

Piyapas Tharavanij, Vasan Siraprapasiri and Kittichai Rajchamaha (2017). Profitability of Candlestick Charting Patterns in the Stock Exchange of Thailand. SAGE Open.

Read the full paper