Explainer · Charts & Analysis
Candlestick charts: what the research says about the patterns
Courses and trading videos treat candlestick patterns as a shortcut to knowing what price does next. If you are a retail trader deciding whether to learn them, what does the evidence actually say? Less than the marketing, and the details matter.

Quick answer
A candle shows four prices for one period: the body spans open and close, the shadows reach the high and low [1]. As signals, the evidence is weak: in a ten-year study of 50 Thai stocks, most reversal patterns had mean returns not statistically different from zero [1].
Key points
- A candle is four prices from one period: open, high, low and close [3].
- Most reversal patterns tested on 50 Thai stocks from 2006 to 2016 earned mean returns not different from zero [1].
- On Dow Jones stocks, whether patterns paid depended on the exit rule, not only the pattern [2].
- A candle shows traded prices, not the price you will get [4].
On this page
What is the least you need to know to read a candle?#
One candle summarises one period, whether a minute or a week. The body runs between the open and the close, and the thin upper and lower shadows (often called wicks) mark the high and low within that time frame [1]. Colour only says whether the close finished above or below the open, and it is a display choice: the Thai study used white for up and black for down [1], while many apps use green and red.
- Prices
- open $100.00, high $104.50, low $98.20, close $103.10invented worked example
- Body
- $3.10 upclose minus open, calculated
- Shadows
- $1.40 above, $1.80 belowcalculated
- Body share of range
- 49.2%$3.10 / $6.30, calculated
That is all the reading you need for this page. For a full walkthrough of bodies, wicks, timeframes and the daily close, go deeper on Chartora's guide to reading candlestick charts.
Have candlestick patterns been tested properly?#
Yes, and the results are a long way from the confidence of a typical pattern cheat sheet. The 2017 SAGE Open study tested one-, two- and three-day bullish and bearish reversal patterns on the 50 largest Thai stocks from July 3, 2006 to June 30, 2016. The mean returns of most patterns were not statistically different from zero, and even the significant ones carried high risk measured by standard deviation [1].
A 2015 study of Dow Jones stocks found eight three-day patterns profitable with one holding rule but not with another [2]. In other words, the exit decision did much of the work. A 2021 PLOS ONE paper reported a deep-learning model guessing direction from candlestick images with 80.82% accuracy on the Taiwan index and 66.53% on the Nikkei 225 [3]: a gap of 14.29 percentage points across different test years (calculated), and accuracy at guessing direction is not trading profit after costs.
| Study | Data tested | What it found |
|---|---|---|
| SAGE Open, 2017 | 50 Thai stocks, daily, 2006 to 2016 | Most reversal patterns: mean returns not different from zero |
| Journal of Banking & Finance, 2015 | Dow Jones stocks | Eight 3-day patterns profitable under one holding rule, not another |
| PLOS ONE, 2021 | Taiwan and Nikkei indexes, daily | Direction accuracy 80.82% on one index, 66.53% on the other |
Sources: [1], [2], [3]. None of these is a test of a retail trader's live results after costs.
Why might a pattern that tested well still fail you?#
Because a backtest and a retail account are different environments. A test works from recorded prices; the SEC notes the last-traded price is not necessarily the price a market order will get [4]. Spreads and fees come on top. And results that vary by market and by exit rule, as above, suggest that a pattern's edge can be fragile. Our page on backtesting pitfalls explains why past tests tend to flatter.
How should a retail trader use candles, then?#
As a compact record of what price did, not as a set of instructions. A few habits keep the reading honest.
- Read the four prices, then the shape
Direction, body, shadows. Name the pattern last, if at all.
- Ask what the candle hides
No volume, no news, and no order of events inside the period. See volume and liquidity for one missing piece.
Candlestick mistakes that cost retail traders#
- Trading a pattern name
Most reversal patterns in the Thai study did not produce returns different from zero [1].
- Ignoring the exit
In the Dow Jones study the holding rule decided whether patterns paid [2].
- Reading the close as your price
The last trade is not a promise of your fill [4].
Frequently asked questions#
Are candlestick patterns reliable?
Not on their own, according to the studies we cite. Results depend on the market, the period and the exit rule [2].
What is the difference between a wick and a shadow?
None. Shadow is the term the research papers use [1]; wick is the common platform word for the same line.
Does a deep-learning model prove candles can predict prices?
No. The 2021 paper measured accuracy at guessing direction on two indexes, with very different scores [3]. That is not the same as profit after costs.
The bottom line#
Learn to read a candle in a minute, then be sceptical of anyone selling patterns as signals. The peer-reviewed tests we cite found weak, fragile results that depend on the market and the exit rule, and none of them includes your costs. Look up the short definition in our candlestick glossary entry and keep the risk disclosure in mind.
Sources
- Profitability of Candlestick Charting Patterns in the Stock Exchange of Thailand (Piyapas Tharavanij, Vasan Siraprapasiri, Kittichai Rajchamaha), SAGE Open, October-December 2017.
- Trend definition or holding strategy: What determines the profitability of candlestick charting?" (abstract, Journal of Banking & Finance 61(C), 2015, pp. 172-183).
- DPP: Deep predictor for price movement from candlestick charts.
- Types of Orders" (Investor.gov).
Education only. This page is not investment, tax or legal advice. Trading and crypto can lose you money. See our risk disclosure.


