How to read candlestick charts: a guide with real data

A candlestick chart draws each trading session as a single shape that shows four prices at once: where the session opened, the highest and lowest prices it reached, and where it closed. Lined up one after another, those shapes show not only where a price went, but how it got there: whether buyers pushed all session, whether sellers took back an early gain, or whether nobody could make any progress.

Candlesticks grew out of the rice markets of 18th-century Japan, where tradition credits the merchant Munehisa Homma with studying them. They reached Western traders mainly through Steve Nison's 1991 book Japanese Candlestick Charting Techniques, and today they are the default chart type on almost every trading platform, Finabase included.

This guide teaches you to read one candle, then a sequence of them. It explains how the Finabase scanner decides whether a body is "long" or a shadow is "short", the four families of patterns in this library, and what our own test of more than 57,000 detected patterns says about how much any single pattern can tell you.

Anatomy of a candle

Each candle has two parts.

The real body is the thick rectangle between the open and the close. It is the part of the session that "counted": the distance between where trading started and where it settled.

• When the close is above the open, the body is drawn green (or hollow on black-and-white charts). Buyers finished the session ahead.
• When the close is below the open, the body is red (or filled). Sellers finished ahead.

The shadows, also called wicks or tails, are the thin lines above and below the body. The upper shadow runs from the top of the body to the session's high; the lower shadow runs from the bottom of the body to the low. Shadows record prices that traded but did not hold.

A common beginner mistake is to read color as "up day" or "down day". It is not. A stock can gap up at the open, drift lower all session and still close above the previous day's close. That candle is red, because it closed below its own open, even though the stock gained on the day. Color compares the close with the same session's open, never with yesterday.

What the proportions tell you

The relationship between body and shadows is where candles become informative.

A long lower shadow is sometimes described as "rejection of lower prices", and a long upper shadow as "rejection of higher prices". Those phrases are a useful shorthand as long as you remember that one session's rejection is a single data point.

Reading a real chart

On a real chart, read candles in three passes:

1. The trend. Ignore individual candles and look at the sequence. Are highs and lows rising, falling or flat? Most candlestick patterns are defined relative to the trend that came before them.
2. The outliers. Look for candles that are unusual for this stock: a body much longer than its neighbors, a shadow that reaches far beyond the recent range, a gap between one candle and the next. Normal candles carry little information. Unusual ones mark the sessions where something changed.
3. The location. Is the unusual candle at a level that already mattered, such as a prior high or low, a moving average or a round number? A long lower shadow that bounces off a support level says more than the same shadow in the middle of nowhere.

The timeframe matters too. A daily candle summarizes a whole session. A weekly candle summarizes five, and a 5-minute candle summarizes a few hundred trades. The shapes read the same way on every timeframe, but a pattern on a weekly chart reflects far more trading than one on an intraday chart.

How the Finabase scanner measures "long" and "short"

Candlestick books describe bodies as "long" or "small" and shadows as "long" or "short". A computer needs numbers. Finabase detects patterns with TA-Lib, the open-source technical analysis library behind many charting platforms, and TA-Lib makes those words precise by comparing each candle with the ones just before it:

Two consequences are worth knowing before you read any pattern page.

Everything is relative to the stock's own recent behavior. A 2% body is long for a utility in a quiet week and short for a volatile chip stock. That is a strength: the same rules work across very different securities.

"Long shadow" is a loose test. It compares the shadow with the candle's own body, not with other candles. For a candle with a tiny body almost any shadow qualifies, which is why some TA-Lib patterns, such as the long-legged doji, fire far more often than the textbook description suggests.

TA-Lib also mostly tests shape, not trend. Most of its detectors check only the candles inside the pattern. Whether the market was actually falling before a bullish reversal pattern is a judgment the detector leaves to you.

The four families of patterns

This library groups the 61 patterns Finabase can detect by how many candles they need.

• Single-candle patterns read one session: dojis, hammers, shooting stars, marubozus and spinning tops. They are the most common and, individually, the weakest evidence. The Hammer candlestick pattern: meaning, rules and reliability page is the model for the rest.
• Two-candle patterns compare a session with the one before it: Engulfing candlestick pattern: bullish and bearish rules and data, Harami candlestick pattern: the inside body and what it signals, Piercing line pattern: how it forms, TA-Lib rules and track record and dark cloud cover, and the neck lines such as On-neck candlestick pattern: when the old low becomes a ceiling.
• Three-candle patterns tell a short story with a beginning, a turn and a confirmation: morning and evening stars, three white soldiers, three black crows.
• Multi-candle patterns span four or five sessions: breakaways, rising and falling three methods and the Hikkake pattern: trading the false breakout of an inside bar.

Patterns are also described as reversal patterns, which suggest the prior trend may end, or continuation patterns, which suggest a pause before the trend resumes. TA-Lib reports each detection as +100 (bullish) or −100 (bearish), and a few patterns use ±80 for a weaker variant or ±200 for a confirmed one.

What 57,000 detected patterns did next

How much does a candlestick pattern actually tell you? We ran all 61 TA-Lib detectors that Finabase uses over two years of daily bars, October 30, 2024 to October 29, 2026, for 60 widely held US stocks and ETFs, from Apple and Microsoft to the SPY, QQQ, GLD and TLT funds. That is 30,060 daily candles and 57,087 pattern detections.

For every detection we asked a simple question: five trading sessions later, had the price moved in the direction the pattern implies? Then we compared that hit rate with the same question asked of every day in the sample, the baseline. Over this period 53.1% of all days were followed by a higher close five sessions later, and 46.7% by a lower one.

Four findings stand out.

Patterns are everywhere. 77.9% of all daily candles carried at least one TA-Lib pattern. When almost every candle is "a pattern", the label alone cannot be special.

Most patterns did no better than a random day. Of the 43 pattern sides with at least 100 cases, 40 sat inside the noise band, the range of hit rates you would expect from the same number of randomly chosen days. Only three fell outside it, and with 43 comparisons, two or three outliers are roughly what chance alone produces.

The outliers were mixed. Three inside up was the strongest, at 61.9% higher after five sessions against the 53.1% baseline, but on only 244 cases. Bearish harami came in 3.2 points above its baseline. The famous bullish engulfing pattern did worse than a random day: 48.5% of 1,055 cases were higher five sessions later.

Eleven patterns never fired at all. Five, like the concealing baby swallow and the mat hold, are simply rare. The other six, including the morning and evening stars, cannot fire in Finabase because of how the app configures them, as explained below.

Each pattern page in this library repeats the test for that pattern, with its own baseline, its confirmed cases and real examples that worked and failed.

Using candlesticks well

Practitioners who get value from candlesticks tend to use them the same way.

• Start with the trend and the level. Decide where a reversal would matter before you look for one. A bullish pattern at long-term support is worth more attention than the same pattern halfway through a decline.
• Wait for confirmation. Most reversal patterns are only potential turns. A common confirmation is the next session closing beyond the pattern's high (bullish) or low (bearish).
• Check volume and momentum. Heavier volume on the pattern day, or an oscillator such as the Relative Strength Index (RSI): how to read and use it turning from an extreme, adds independent evidence.
• Use the pattern to place risk. A pattern's high or low is a natural point where the idea is proven wrong. That is often its most practical use: it tells you where you would be wrong before you act.
• Prefer higher timeframes for big decisions. A weekly pattern reflects five sessions of trading and is less likely to be noise.

How Finabase shows candlesticks and patterns

• Charts. Every symbol chart draws candles by default. Add any of the 61 pattern studies and Finabase marks each candle TA-Lib flags, labelled bullish or bearish.
• The Scanner. The Scanner's candlestick tab lists the stocks where each pattern fired, split into Bullish and Bearish by the sign of TA-Lib's output.
• The Signals tab's candlestick gauge. On a symbol's Signals tab, each of the 61 patterns casts one vote: Buy if its most recent detection in the last eight bars was bullish, Sell if bearish, Neutral if it did not fire. The votes combine into a score from Strong Sell to Strong Buy. Like every gauge input, it describes the recent chart. It does not forecast it.

Two details of how Finabase runs TA-Lib affect what you see:

Frequently asked questions

What do the colors on a candlestick chart mean?

Green (or hollow) means the session closed above its own open; red (or filled) means it closed below its open. Color does not tell you whether the price rose from the previous session, so a red candle can still be an up day.

What is the difference between the body and the wick?

The body spans the open and the close, the part of the session's range that held. The wicks, or shadows, extend to the high and low, prices that traded but were given back by the close.

Are candlestick patterns reliable?

On their own, not very. In our test of 60 large US stocks and ETFs, most patterns' five-session hit rates were within the range a random day produces. They are more useful as a way to read a session and to place risk than as standalone signals.

Which timeframe is best for candlesticks?

The rules are the same on every timeframe. Daily and weekly candles summarize more trading and are less noisy than intraday candles, which is why most pattern research, including ours, uses daily bars.

Do I need to memorize all 61 patterns?

No. If you understand what a long body, a small body and a long shadow mean, and you always ask where the candle appears in the trend, you can read most patterns from first principles. The pattern names are a shared vocabulary, not a requirement.

Summary

A candlestick packs a session's open, high, low and close into one shape, and the balance between its body and shadows shows who controlled the session. Reading candlesticks well means looking at the trend first, then at the candles that are unusual for that stock, then at where they appear. The Finabase scanner judges "long" and "short" the way TA-Lib does, against each stock's own recent candles, and it mostly leaves the trend to you. Our test of 57,087 detected patterns found that the shapes alone carried little edge over a random day. Use candlesticks to understand the market's behavior and to define your risk, and look for agreement from the trend, the level, volume and momentum before you trust any single pattern.