How to Read a Candlestick Chart
To read a candlestick chart, start with the four prices each bar records: open, high, low and close. The body shows the distance between open and close, the wicks show prices that were reached and rejected, and the meaning of any candle depends entirely on where it appears.
A candlestick is four numbers drawn as a shape. Almost everything difficult about reading them comes from people learning the shapes before learning what the four numbers mean.
Before you start
A bare chart with no indicators on it. Every indicator is a transformation of these same four prices, so learning them first is learning the input rather than the output.
One instrument you will look at repeatedly. Bar sizes differ enormously between markets, and “large” only means anything relative to a specific one.
The median bar height of that instrument over the last fifty bars. One number, written down. It is what every subsequent judgement about size is measured against.
The steps
1. Learn the four prices before any pattern
Open, high, low, close, over one fixed period. The colour is just whether the close was above or below the open. Nothing else is recorded.
2. Read the body as the argument and the wick as the rejection
The body spans open to close and shows where the period settled. The wicks show prices that were reached and then given back.
3. Look at where the candle is before what it is
At a level, after a long move, or in the middle of nothing. Location changes the meaning completely, and it is the half most explanations skip.
4. Measure the bar against its neighbours
Compare it to the median you wrote down. A “large” bar is only large relative to what this instrument normally does, and that changes over time.
5. Give more weight to the close than to the shape
The close is the price the period agreed on and the one most systems and traders reference. A dramatic wick with a close back in the middle resolved nothing.
6. Check whether anyone was participating
A large bar on very little volume is a thin market moving, not a decision. The histogram is the second half of the same story.
7. Treat a named pattern as a description, not a prediction
The names are shorthand for shapes that have already formed. They compress communication; they do not carry a base rate on their own.
How to tell it worked
Test yourself on 20 bars, using the right-hand edge honestly.
Cover the chart to the right of a bar and state what the four prices were. 20 out of 20 correct means the anatomy is genuinely learned rather than recognised. This sounds trivial and most people cannot do it at first.
Then classify each of those 20 bars as larger or smaller than your median. If your instinct disagrees with the measurement more than a few times, you are reading size by drama rather than by scale.
Finally, for any bar you would have acted on, write down the level it was at. A bar you would trade that sits at no level is a shape being read in isolation, which is the error step three exists to prevent.
Why scale has to be measured
Bar sizes vary enormously within a single series. On the shared price history the median bar spans 0.493, the tenth percentile 0.17 and the ninetieth 1.101 — a 6.5-fold spread. The smallest bar measured 0.022 and the largest 2.338, a difference of more than a hundred times.
So “a big candle” is not a fixed thing. A bar that is dramatic during a quiet stretch is unremarkable during a volatile one, and reading them the same way guarantees inconsistent decisions.
Which is what the median in the prerequisites is for. One number, updated occasionally, converts every size judgement from an impression into a comparison.
The three candles worth knowing, and what they record
A long lower wick records a price that was reached and refused. Sellers pushed down, the period closed well above the low, and the low did not hold. That is all it says. Whether it matters depends on whether the low was somewhere anyone was watching.
A bar with almost no body records disagreement. Open and close in nearly the same place after a period of movement means neither side finished ahead. In the middle of a range that is unremarkable; after a long directional run it is the first period that did not continue.
A bar much larger than its neighbours records a decision. Measured against the median you wrote down, it is a period where something was resolved rather than debated — and the close tells you which way. Acting on any of the three costs a round trip, which on the shared series is 2% of a median bar’s range, so a candle is worth acting on only when it sits somewhere that already mattered.
The original data
Of the 24,971 unique videos in research/search-study-corpus.jsonl, 101 have an instruction-shaped
title mentioning candlesticks, at a median of 19,243 views across 86 channels, with a maximum of
5,378,610. Chart patterns more broadly appear in 69 at a median of 4,718. The counts come from
site/rank_howto.py.
101 videos at a 19,243 median against 69 chart-pattern videos at 4,718. The more fundamental subject outperforms the more advanced-sounding one by four times per video, which is consistent across this corpus: the foundational topics are where the audience actually is.
The answer to the question on that chart is that the wick alone does not say. A long wick at a level you marked beforehand is information; the same wick in open space is a period that moved and came back. The shape is identical in both cases, which is precisely why step three puts location ahead of pattern.
When it fails
A range produces an unending stream of convincing reversal candles. Price turns at the top, prints a long upper wick, turns at the bottom, prints a long lower wick, and does it again. Every one is a textbook example and none of them leads anywhere, because the turning is what a range does rather than evidence about what comes next. Pattern-based reading is at its least reliable exactly where it generates the most signals.
The second failure is learning names before anatomy. The names describe shapes whose components have not been understood.
A third is judging size without a reference. A 6.5-fold spread makes “big” meaningless on its own.
A fourth is reading a candle in open space. There is nothing for it to be evidence about.
A fifth is ignoring the close. The wick is the story people tell; the close is what happened.
And a sixth is treating a gap as a candle. It is four prices with no trading in between them.
Related
Candlesticks covers the full anatomy and the named patterns in detail. Price action is reading the chart without indicators, which this is the foundation of. And support and resistance supplies the locations that make any individual candle mean something.
I spent far too long learning pattern names and far too little time asking where the candle was. A hammer at the bottom of a long decline and a hammer in the middle of a range are the same drawing and completely different information. Once I started reading location first and shape second, most of the pattern vocabulary turned out to be unnecessary.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.