Candlestick Patterns: How They Work and What the Research Says

Important note: this article is educational and is not financial advice. It explains what candlestick charts and patterns are, where they came from, and what academic research has found about them. Nothing here is a recommendation to buy, sell, or hold any security, currency, or commodity. Trading involves real risk of loss. As the research section below shows, the evidence that candlestick patterns can be used to beat the market after costs is weak and mixed.

A candlestick chart is a way of drawing price movements in a market. Each "candle" summarizes the trading in one period, such as a day, an hour, or a minute, using four numbers: the opening price, the highest price, the lowest price, and the closing price. A candlestick pattern is a particular shape or sequence of candles that some traders believe signals what prices will do next. Patterns have colorful names, such as hammer, doji, shooting star, engulfing, morning star, and three black crows. They are among the most widely taught tools of technical analysis, the practice of trying to forecast prices from past price and volume data.

This article covers how a candlestick is built, the history of the method, the most common patterns and what traders claim they mean, and how patterns are used in practice. It then looks at what decades of academic testing have found about whether they actually predict anything, the main methodological problems with such tests, and common misconceptions.

How a candlestick is built

The four prices

Every candle is built from four values for its time period:

• Open: the price of the first trade in the period.
• High: the highest price traded during the period.
• Low: the lowest price traded during the period.
• Close: the price of the last trade in the period.

These are often abbreviated OHLC. A traditional bar chart shows exactly the same four values, using a vertical line from high to low with small ticks for the open and close. The candlestick format shows them in a way that makes the relationship between open and close stand out visually.

The real body

The thick rectangle in the middle of a candle is called the real body, or simply the body. It spans the range between the open and the close.

• If the close is higher than the open, the price rose during the period. The body is traditionally drawn hollow (white with a black outline), or in modern software usually green.
• If the close is lower than the open, the price fell. The body is traditionally filled black, or in modern software usually red.

Software lets users choose their own colors, so conventions vary. A trader reading someone else's chart should check which color means what.

The shadows

Thin lines extending above and below the body are called shadows, also called wicks or tails. The upper shadow runs from the top of the body to the high of the period. The lower shadow runs from the bottom of the body to the low. A candle may have no shadow on one side or both, if the open or close was also the high or low.

The size of the body relative to the shadows is what gives each candle its character. A long body with short shadows shows that the price moved steadily in one direction. A small body with long shadows shows that the price moved a great deal during the period but ended up close to where it started.

A worked example of building candles

The following numbers are hypothetical, chosen only to show how candles are drawn. They do not describe any real security.

On Day 1, the price opened at 50 and closed at 52, so the body is hollow or green and runs from 50 to 52. The shadows reach up to 52.50 and down to 49.50. On Day 2, the price fell from 52 to 48.50, giving a long filled or red body. On Day 3, the price dropped as low as 45 during the day but recovered to close slightly above its open. The result is a tiny body at the top of the day's range with a very long lower shadow. Traders would call a candle shaped like Day 3 a "hammer," discussed below.

Timeframes

The same method works for any period length. A daily chart has one candle per trading day, a weekly chart one per week, and an intraday chart might have one candle per five minutes. A pattern on a weekly chart summarizes far more trading than the same pattern on a one-minute chart. Pattern guides often suggest that signals on longer timeframes carry more weight, though, as discussed below, that claim is part of the trading literature rather than a well-tested finding.

Variations

Several variants exist. Hollow candlesticks use two dimensions at once. Whether a body is filled or hollow shows whether the close was below or above the open, and its color shows whether the close was below or above the previous close. Heikin-Ashi candles ("average bar" in Japanese) are calculated from averages rather than raw prices. Each Heikin-Ashi close is the average of the real open, high, low, and close, and each Heikin-Ashi open is the average of the previous Heikin-Ashi open and close. This smooths out the chart and makes trends easier to see, but the bodies no longer match actual opening and closing prices. Some software also varies the width of candles to show trading volume.

History

Rice trading in Japan

Candlestick charting is associated with the rice markets of Tokugawa-era Japan. The Dōjima Rice Exchange in Osaka received a license from the shogunate in 1697 and became the center of Japan's rice brokerage system. Trading in rice coupons, promises of future delivery, emerged around 1710. This is often described as one of the earliest organized futures markets in the world. The exchange was later officially sponsored and regulated by the shogunate from 1773, and was formally dissolved in 1939.

Munehisa Homma

Much of the credit for candlestick analysis traditionally goes to Munehisa Homma (also written Honma), a rice merchant from Sakata who lived from 1724 to 1803 and traded at Dōjima. He is credited with a 1755 book, San-en Kinsen Hiroku, translated as The Fountain of Gold: The Three Monkey Record of Money, which discusses the emotional and psychological side of trading. Other books, including works associated with "Sakata" trading rules, are attributed to him by some sources, but those attributions are uncertain. Popular trading sites often repeat dramatic claims about his fortune and his winning streaks. The Wikipedia figure of about $10 billion in today's money has no supporting citation, and such stories are best treated as legend.

Did Homma actually use candlesticks?

The historical record is less clear than the popular story. Steve Nison, the trader who did most to introduce candlesticks to Western readers, has argued that Homma probably did not use candlestick charts himself. In Nison's account, the charts as we know them likely emerged later, in the second half of the 1800s, around the early Meiji period. Homma's lasting contribution, on this view, was his writing on market psychology, and his name later became attached to the charting method. The honest conclusion is that candlestick charts are a Japanese invention of the 18th or 19th century, connected to the rice trade, and that their exact origin is not well documented.

Arrival in the West

Candlestick charts were little known outside Japan until Steve Nison's Japanese Candlestick Charting Techniques, first published in 1991, which popularized them among Western traders. Nison followed it with Beyond Candlesticks in 1994. Other widely used guides include Gregory Morris's Candlestick Charting Explained. Once charting software made candlesticks a standard display option, they quickly became the default chart type on many trading platforms. Today many retail traders see price data in candlestick form whether or not they use the patterns.

Why traders think patterns mean something

The theory behind candlestick patterns is a theory about crowd psychology. A candle records a struggle between buyers and sellers during one period. Its shape is read as a summary of who won that struggle and how decisively.

• A long hollow or green body says buyers were in control from open to close.
• A long filled or red body says sellers were in control.
• A long lower shadow says sellers pushed the price down during the period, but buyers pushed it back up before the close.
• A long upper shadow says buyers pushed the price up, but sellers drove it back down.
• A very small body says neither side won: the period ended close to where it began.

Pattern guides then add a second ingredient, which is context. The same candle shape is said to mean different things depending on what came before it. A small-bodied candle with a long lower shadow after a long decline is read as a sign that sellers are losing their grip. The same shape after a long rise is read as a warning. This is why many patterns come in pairs with different names for the same shape, such as the hammer and the hanging man.

Most guides also stress confirmation. A pattern is not to be acted on alone, but only if the next candle or candles move in the expected direction, or if other evidence such as volume, support and resistance levels, or other indicators agrees.

Common patterns

The Wikipedia article on candlestick patterns, citing a trading education site, says there are 42 recognized patterns, divided into simple patterns (one candle) and complex patterns (several candles). Different books list different numbers. The descriptions below report what pattern guides claim each pattern means. They are not evidence that the claims are true. That question is covered in the research section.

Single-candle patterns

Two-candle patterns

Three-candle and longer patterns

Reading a pattern: a worked example

Return to the hypothetical three-day table above. Suppose the hypothetical stock had been falling for several weeks before Day 1. Day 2 is a long red candle, continuing the decline. Day 3 is a hammer: the price fell to 45, but buyers pushed it back to close near the top of the day's range. A pattern guide would say that, in the context of a downtrend, the hammer suggests sellers may be exhausted. It would then usually say to wait for confirmation, such as Day 4 opening and closing above Day 3's close, before treating the pattern as a buy signal.

Notice how many judgment calls are involved even in this simple case. How long must the downtrend be? How long must the lower shadow be relative to the body? Two times? Three? What counts as "near the top of the range"? How strong must the confirming candle be? Different books answer these questions differently, and as Wikipedia notes, pattern recognition is subjective. Charting software that detects patterns automatically has to impose its own fixed rules. That subjectivity turns out to be a central problem in testing whether patterns work.

How patterns are used in practice

Candlestick patterns are used mostly by short-term traders in stocks, futures, foreign exchange, and cryptocurrencies. A few typical uses:

• Entry and exit timing. A trader who has decided on other grounds to buy might wait for a bullish pattern before entering, or exit when a bearish pattern appears.
• Combination with other tools. Patterns are often combined with support and resistance levels, moving averages, or volume. A hammer at a long-standing support level is usually treated as more meaningful than one in the middle of nowhere.
• Visual summary. Many people use candlestick charts simply because they make the open-close relationship easy to see, without trading on named patterns at all.
• Automated screening. Software can scan thousands of instruments for defined patterns. This has also made it easier for researchers to test patterns systematically.

From human eyes to rules in code

To scan for a pattern by computer, every vague phrase in a pattern guide has to become a number. Here is a hypothetical rule set for a hammer, written only to show what such a translation looks like. It is not a recommended or tested definition:

1. The lower shadow is at least twice the length of the body.
2. The upper shadow is no more than 10 percent of the candle's full range.
3. The body sits in the top third of the range.
4. The close is below its average close of the previous ten periods, as a rough stand-in for "after a decline."

Each of these thresholds is a choice. Change "twice" to "three times," or ten periods to twenty, and the program will flag a different set of candles. This is why two software packages can disagree about whether a hammer formed on a given day. It also explains why researchers who test patterns must state their exact definitions, and why a pattern can look effective under one definition and useless under another. The difficulty runs in both directions. A trader reading charts by eye is applying definitions too, just unstated and inconsistent ones, which makes their own track record hard to evaluate honestly.

Newer research, such as the similarity-matching study of Tao and colleagues mentioned below, avoids the named patterns altogether. Instead it lets algorithms search past data for candle shapes that were followed by particular price moves. This removes some of the subjectivity, but it greatly increases the number of possibilities being searched, and with it the risk of finding patterns that are only noise.

What the research says about predictive power

Candlestick patterns have been tested many times by academic researchers. As Wikipedia summarizes, academic opinion is divided on whether these patterns have predictive power. The overall weight of evidence for large, liquid stock markets is that most patterns do not reliably predict returns after realistic costs, although some studies, mostly in particular markets and periods, have found exceptions.

The broader question: can past prices predict future prices?

Candlestick research sits inside a larger debate about technical analysis. The weak form of the efficient market hypothesis holds that current prices already reflect all information contained in past prices. If it is true, no rule based only on past prices should produce consistent profits beyond what is earned by taking risk, after costs.

A major review by Cheol-Ho Park and Scott Irwin, published in the Journal of Economic Surveys in 2007, examined 95 modern studies of technical trading rules of all kinds, not only candlesticks. Of these, 56 reported positive results, 20 negative results, and 19 mixed results. The authors concluded that technical strategies had generated economic profits in a variety of speculative markets, at least until the early 1990s. But they also warned that most of the studies suffered from testing problems, including data snooping, choosing rules after looking at the results, and difficulties in estimating risk and transaction costs. In other words, much of the positive evidence may be weaker than it looks.

Evidence in favor

Caginalp and Laurent (1998). G. Caginalp and H. Laurent, in "The Predictive Power of Price Patterns" in Applied Mathematical Finance, tested three-day candlestick reversal patterns on daily open, high, low, and close prices for all S&P 500 stocks from 1992 to 1996. They used standard definitions of the patterns, dropped conditions on size, and applied an essentially non-parametric test. They reported evidence of significant predictive power, and concluded that the results show traders are influenced by price behavior. This is the most frequently cited positive result for candlesticks.

Lu, Shiu, and Liu (2012). Lu, Shiu, and Liu, in the Review of Financial Economics, studied two-day candlestick patterns in the component stocks of the Taiwan Top 50 Tracker Fund from October 2002 to December 2008. Their strategy bought on a bullish pattern and held until a bearish pattern appeared, or the reverse. They examined three bullish and three bearish reversal patterns and found that the three bullish patterns were profitable in the Taiwan market. They checked the results with out-of-sample tests, different market conditions, and a bootstrap method.

Lu (2014). In the Pacific-Basin Finance Journal, Tsung-Hsun Lu tested single-candle patterns on daily Taiwan stock data from January 1992 to December 2009. He used a systematic classification based on the relative positions of the four prices, and reported that four patterns were profitable for the Taiwan stock market after transaction costs. He also ran bootstrap, out-of-sample, and sub-period checks.

Pattern similarity methods. A 2017 study by Tao and colleagues in Mathematical Problems in Engineering, cited by Wikipedia, analyzed the predictive power of "K-line" patterns using similarity matching and clustering. This represents a newer line of work that tries to find predictive shapes by computer, rather than by testing the traditional named patterns.

Evidence against

Marshall, Young, and Rose (2006). Ben Marshall, Martin Young, and Lawrence Rose, in "Candlestick Technical Trading Strategies: Can They Create Value for Investors?" in the Journal of Banking and Finance, tested candlestick patterns on the component stocks of the Dow Jones Industrial Average from 1992 to 2002. A related paper by the same authors, "Market Timing with Candlestick Technical Analysis," tested 28 patterns based on one to three days of prices, using a 10-day holding period. Under tests that assumed returns were normally distributed, none of the bullish single-day signals reliably predicted positive abnormal returns, one predicted negative returns, and two bearish multi-day signals were followed by reliably positive returns, the opposite of what they are supposed to signal. Under bootstrap tests, which do not assume normality, there was no evidence that any of the signals predicted price movements. Marshall's doctoral thesis on the topic concluded that there is no evidence a trader following candlestick analysis would outperform the market.

Marshall, Young, and Cahan (2008). The same research group then asked whether candlesticks work better in their country of origin. In "Are Candlestick Technical Trading Strategies Profitable in the Japanese Equity Market?", published in the Review of Quantitative Finance and Accounting, Ben Marshall, Martin Young, and Rochester Cahan tested Japanese stocks from 1975 to 2004. A summary of the paper reports that candlestick strategies were not profitable over the full 30 years, in any of three 10-year sub-periods, or in bull or bear markets.

Horton (2009). Marshall Horton's "Stars, Crows, and Doji: The Use of Candlesticks in Stock Selection," in The Quarterly Review of Economics and Finance, examined candlestick methods on 349 stocks. Using more data and alternative tests, the study contradicted an earlier article in the literature. It found little value in candlesticks and more support for the weak form of the efficient market hypothesis.

Western chart patterns

It is worth distinguishing candlestick patterns from Western chart patterns such as head-and-shoulders or double bottoms. In a well-known 2000 paper in the Journal of Finance, Andrew Lo, Harry Mamaysky, and Jiang Wang developed an automatic, statistical method for recognizing such patterns using kernel regression. They applied it to a large sample of U.S. stocks from 1962 to 1996. They found that several technical patterns did provide incremental information, meaning that returns after a pattern had a somewhat different distribution than returns in general, and might have some practical value. That is a statement about information content, not proof of trading profits after costs. It also concerns Western chart formations, not candlestick patterns. It is often cited, wrongly, as if it validated candlesticks.

Summary of studies

How to read the conflicting results

Several patterns in the evidence are worth noticing.

• Large, heavily traded U.S. stocks and the Japanese market show little or no edge. These are among the most liquid and closely watched markets in the world, where any easy edge would be expected to be competed away.
• Positive findings tend to come from specific markets, periods, and pattern definitions. That does not make them wrong. It may reflect real differences between markets, such as the share of retail traders. But it raises the question of whether the results would hold in new data.
• Even the more favorable study of U.S. stocks used a short window. Caginalp and Laurent's data covered five years. Later work by Marshall and colleagues, using longer data and bootstrap methods, did not find similar value in large U.S. stocks.
• Statistical method matters. Several signals that looked significant under normal-distribution assumptions disappeared under bootstrap tests, which better reflect the real, heavy-tailed behavior of stock returns.

A balanced summary is this: there is no consistent evidence that traditional candlestick patterns, used on their own, reliably predict returns in major stock markets after costs. A few studies have found exceptions in particular markets and periods, which may or may not persist.

Why testing patterns is hard

Testing candlestick patterns properly runs into problems that apply to almost all technical analysis research.

Definition ambiguity

Pattern definitions are verbal and vague: "long" shadow, "small" body, "near" the high, "after a downtrend." To test a pattern, a researcher must turn these words into exact rules. Different rules produce different results. Supporters of candlestick analysis can always argue that a failed test used the wrong definition. One reader's response to the Marshall study, for instance, argued that it ignored the short-term trend context that patterns depend on. This flexibility makes the method hard to falsify.

Data snooping

With dozens of patterns, many possible definitions of each, several holding periods, and many markets, a researcher can run thousands of tests. By chance alone, some will look significant. If only the successful combinations are reported or used, they will appear to work historically and then disappoint in practice. Park and Irwin identified data snooping as one of the main weaknesses of the technical analysis literature. Out-of-sample testing, in which rules chosen on one period are tested on a separate later period, is the main defense.

Transaction costs

Candlestick strategies tend to trade often, because they react to short-term signals. Each trade costs money in commissions, the bid-ask spread, and the price impact of the order. A small predictive edge before costs can disappear entirely after them. Studies that report results only before costs, or assume unrealistically low costs, overstate profitability.

Risk adjustment

A strategy that earns higher returns by taking more risk has not necessarily found an edge. Proper tests compare returns to an appropriate benchmark and adjust for the risk taken. That is harder than it sounds, and different methods give different answers.

Look-ahead bias and survivorship bias

Backtests must use only information that would have been available at the time. For example, a pattern that is only complete at the close of a day cannot be traded at that day's open. Tests that use today's list of index members to study the past also exclude companies that failed or were removed, which can make historical results look better than they were.

Self-fulfilling and self-defeating effects

Because candlestick patterns are widely taught and displayed by default on many platforms, some argue that patterns can work briefly because many traders act on them at once. Others argue the opposite: any widely known edge will be exploited and disappear. Neither effect has been clearly established for candlesticks, and both make historical results a weak guide to future behavior.

Common misconceptions

"Candlestick patterns are proven to predict prices." They are not. Several careful studies of major markets found no reliable predictive value, while a minority found value in specific markets and periods.

"The patterns are centuries old and were used to make a famous fortune." Homma's fortune is a matter of legend, the numbers quoted online are unsupported, and Steve Nison himself has argued that candlestick charts probably appeared after Homma's time.

"A candlestick chart contains more information than a bar chart." It does not. Both show the same four prices. The candlestick format makes the open-close relationship easier to see, which is a difference in presentation, not data.

"If a pattern worked in a backtest, it will work going forward." Backtests are vulnerable to data snooping, look-ahead bias, and underestimated costs. Results that hold only in the data used to discover them often fail in new data.

"Green means the price went up from yesterday." In standard candlesticks, color shows the relationship between the open and close of the same period, not the change from the previous period. Hollow-candle variants add a second color rule for the previous close, which can cause confusion.

"Patterns work the same on any timeframe and in any market." The research finds different results in different markets and periods. There is no strong evidence that any pattern works universally.

"Lo, Mamaysky, and Wang proved technical analysis works." Their paper found that some Western chart patterns carried incremental statistical information. It did not test candlestick patterns and did not show profits after costs.

Frequently asked questions

What is a candlestick in a price chart?

A visual summary of trading in one time period, showing the open, high, low, and close. The body spans the open and close, and the shadows show the high and low.

What does a doji mean?

A doji is a candle whose open and close are nearly equal, so the body is a thin line. Pattern guides read it as a sign of indecision between buyers and sellers, possibly signaling a turning point when it follows a strong trend. Whether this reliably predicts anything is not established by research.

What is the difference between a hammer and a hanging man?

They have the same shape: a small body near the top of the range and a long lower shadow. A hammer appears after a decline and is read as bullish. A hanging man appears after a rise and is read as bearish. The difference is purely the context that comes before the candle.

Who invented candlestick charts?

They developed in Japan in connection with rice trading. They are traditionally credited to Munehisa Homma (1724 to 1803), but Steve Nison has argued that the charts probably came into use later in the 19th century. Nison's 1991 book introduced them widely in the West.

Do candlestick patterns work?

The research evidence is mixed and mostly negative for major markets. Studies of U.S. Dow Jones stocks (1992 to 2002), Japanese stocks (1975 to 2004), and a sample of 349 stocks found little or no value. Studies of S&P 500 stocks (1992 to 1996) and of Taiwan's market found some patterns with predictive power or profits. No study has established that the patterns reliably beat the market in general after costs.

Why do so many traders use them if the evidence is weak?

Several reasons are plausible. Candlestick charts are a clear way to display price data even without trading on patterns. Patterns give traders a structured way to describe what they see. People are prone to remembering hits and forgetting misses. Many trading platforms show candlesticks by default, and much trading education teaches them. Popularity does not establish effectiveness.

Are candlestick patterns useful in cryptocurrency or forex markets?

They are widely used in both. The academic studies summarized in this article mostly concern stock markets, so they do not directly answer the question for other markets. The same methodological concerns, especially data snooping and transaction costs, apply to any market.

How should a beginner think about candlestick patterns?

As a way of reading and describing price action, not as a proven forecasting system. Learning to read candles helps in understanding what happened in a market during a period. Treating patterns as reliable buy or sell signals is not supported by the weight of the research. Anyone considering trading should understand the risks and, if appropriate, consult a qualified, independent professional. This article is educational and not financial advice.

Summary

Candlestick charts are a compact way of drawing the open, high, low, and close of each trading period. The body shows the move from open to close and the shadows show the extremes. They grew out of Japan's rice markets and came to the West through Steve Nison's work in the early 1990s. Today they are the default chart type for many traders. Candlestick patterns, such as hammers, dojis, engulfing patterns, and morning stars, are traditional interpretations of what particular shapes say about the balance between buyers and sellers.

As a way of seeing market behavior, candlesticks are useful and harmless. As a forecasting tool, the evidence is weak. Careful tests in large U.S. stocks and in Japan, the method's home market, found no reliable value. Positive findings exist but are concentrated in particular markets, periods, and pattern definitions. Like most technical analysis, pattern testing is vulnerable to data snooping, vague definitions, and the erosion of small edges by trading costs. The most accurate statement is that candlestick patterns are a widely used language for describing price action, not a demonstrated way to beat the market.

Sources

• Candlestick pattern - Wikipedia
• Candlestick chart - Wikipedia
• Honma Munehisa - Wikipedia
• Dōjima Rice Exchange - Wikipedia
• Caginalp, G. and Laurent, H. (1998). The predictive power of price patterns. Applied Mathematical Finance 5(3-4) - RePEc
• Marshall, B., Young, M. and Rose, L. (2006). Candlestick technical trading strategies: Can they create value for investors? Journal of Banking and Finance 30(8) - EconPapers
• Candlesticks, Fiddlesticks? (summary of Marshall, Young and Rose, Market Timing with Candlestick Technical Analysis) - CXO Advisory
• Marshall, B., Young, M. and Cahan, R. (2008). Are candlestick technical trading strategies profitable in the Japanese equity market? Review of Quantitative Finance and Accounting 31(2) - RePEc
• Horton, M. J. (2009). Stars, crows, and doji: The use of candlesticks in stock selection. The Quarterly Review of Economics and Finance 49(2) - RePEc
• Lu, T.-H., Shiu, Y.-M. and Liu, T.-C. (2012). Profitable candlestick trading strategies: The evidence from a new perspective. Review of Financial Economics 21(2) - RePEc
• Lu, T.-H. (2014). The profitability of candlestick charting in the Taiwan stock market. Pacific-Basin Finance Journal 26 - RePEc
• Park, C.-H. and Irwin, S. H. (2007). What do we know about the profitability of technical analysis? Journal of Economic Surveys 21(4) - RePEc
• Lo, A., Mamaysky, H. and Wang, J. (2000). Foundations of Technical Analysis. Journal of Finance 55(4) - MIT
• Tao, L. et al. (2017). K-Line Patterns' Predictive Power Analysis Using the Methods of Similarity Match and Clustering. Mathematical Problems in Engineering