Candlesticks, and what a candle cannot tell you

The most widely used chart format, and a good example of how presentation can imply more than the data holds.

What a candle is

A candlestick draws chapter 2's four prices for one interval:

• The body spans open to close. Filled or coloured one way if the close was below the open; the other way if above.
• The wicks extend to the high and the low.

That is the entire encoding. Four numbers, drawn so the eye can take in direction and range at once.

It is excellent design. A hundred candles let you see at a glance what a hundred rows of a table would not. But reading a candle gives you exactly the four numbers and nothing further — no more than the table, better arranged.

The vocabulary

Particular shapes have names: a long body with no wicks, a tiny body with long wicks, a small body at the top of a long lower wick, and so on. Each name comes with a story about what market participants were feeling — indecision, rejection, exhaustion, capitulation.

Separate the two things being offered.

The shape is data. "Close was near the high and the low was far below" is a factual description of four numbers.

The story is interpretation. "Buyers rejected lower prices and seized control" is a narrative imposed on four numbers, and the four numbers cannot confirm it.

The narrative may be right. It is not what you observed.

The path problem

This is the chapter's central point, and it disposes of most pattern interpretation.

A candle fixes four points. Infinitely many paths pass through them.

Consider a candle that opens at 100, falls to 95, recovers and closes at 104, with a high of 105. The standard story is: sellers pushed it down, buyers absorbed the selling and drove it up through the day — strength.

The identical candle is produced by:

• A fall to 95 in the first two minutes on thin volume, then a slow drift up all day
• A rise to 105 in the morning, a collapse to 95 at midday, a recovery to 104 at the close
• Ten oscillations between 95 and 105 ending at 104
• A flat day at 100 with a single 95 print and a single 105 print on tiny quantities

Four of these describe quite different days. They draw the same candle. The chart cannot distinguish them, and therefore the emotional narrative — who was in control, when, and with what conviction — is not recoverable from the drawing.

Lower the interval and the problem recurs. A one-minute candle has the same issue at a smaller scale, and you can never reach the bottom, because below the tick there is nothing and at the tick you are no longer looking at candles.

Why this matters for testing

The research literature notes that academic work on technical analysis is generally limited to techniques that can be easily expressed in mathematical form, with visual chart patterns tested only more recently via pattern-recognition algorithms.

That asymmetry has a consequence worth stating plainly. Rules expressible in mathematics can be applied mechanically to decades of data and scored. Visual pattern reading cannot, because two experienced readers will disagree about whether a given shape qualifies.

A method whose application depends on the reader's judgement is difficult to falsify — any failure can be attributed to misreading rather than to the method. That is not proof the method is wrong. It does mean it sits outside the kind of evidence chapter 9 is able to report, and claims of its effectiveness rest on testimony rather than measurement.

Working the problem

Two days, identical candles — same open, high, low, close.

Sequence A. The instrument opens, immediately drops to the low on a handful of trades as one seller clears a position, then recovers within ten minutes and trades quietly in a narrow band for the rest of the session, ticking up to close near the high. Essentially an uneventful day with a brief liquidity dislocation at the open.

Sequence B. The instrument opens, rises steadily to the high by midday on heavy volume, then reverses sharply through the afternoon to the low on heavier volume still as a large holder distributes, and in the last fifteen minutes is bid back up to close near the high. A violent, contested session ending with a late buyer.

The candles are identical. Same four numbers.

What that implies about reading intent:

Intent is not in the drawing. Any statement of the form "buyers took control" is an inference about a path the candle does not record. In A there were barely any sellers; in B there was heavy distribution. The candle is silent.

Volume helps and does not solve it. Attaching volume distinguishes a quiet day from a busy one, which separates A from B at the level of total activity — but it still does not say when within the session the activity occurred, and that is what the narrative claims to know.

The fix is more granular data, not more interpretation. Intraday prints, or at minimum a lower interval, recover some of the path. Staring harder at the candle recovers nothing, because the information was never encoded.

The reasonable use, consistent with how this subject is written: treat a candle as a compact display of four numbers that lets you see range and direction quickly. That is genuinely useful. Treat the named patterns as vocabulary for describing shapes, not as evidence about what anyone was thinking.

The point

A candlestick encodes open, high, low and close, and nothing else — excellent visual design that conveys the same information as four table columns, better arranged. The named patterns bundle a factual shape with an emotional narrative, and only the shape was observed. Infinitely many intraday paths produce the same four numbers, so a quiet day with one stray print and a violently contested session can draw the identical candle, which means intent cannot be read from the drawing. Because visual patterns depend on the reader's judgement, they are also hard to test mechanically, which is why the evidence in chapter 9 concerns mathematical rules rather than chart reading.