Market Structure: Reading Trend and Range from Swing Points

Market structure is usually introduced in one sentence — "higher highs and higher lows" — and the reader is then left to apply it to a chart that looks nothing like the example. The trouble is that the sentence is a conclusion, not a starting point: before it comes a step nobody explains, which is identifying the points being compared at all.

A swing point is a high above its neighbours or a low beneath them, and it is the unit everything else is built from. One rule governs reading it: compare each high with the previous high, and each low with the previous low.

Comparing a high with the low before it is the most common error in structure reading, and a silent one: it produces labels that look orderly and describe nothing. Get this right and the rest follows without memorisation.

Once the points are labelled, the state describes itself without judgement:

• Uptrend — higher highs with higher lows (HH + HL)
• Downtrend — lower highs with lower lows (LH + LL)
• Range — no consistent progression either way

The diagram below shows the reference uptrend on gold, labelled and computed:

Note what the diagram shows and the definition does not: the first two points carry no label, because there is nothing to compare them against. Labelling the first point is guessing rather than reading — a detail that separates disciplined reading from impressionistic reading.

This sequence is also the definition of trend; the two are not different things but two levels of the same description.

Structure's largest practical benefit is not naming the state but defining when the reading becomes wrong — at a price known before entry.

In the uptrend above, the last higher low sits at 2455. A close beneath it breaks the "higher lows" condition and ends the uptrend by definition rather than by opinion. In the mirrored downtrend the condition sits at 2430 — a close above the last lower high:

The value is that the number is known before anything happens. Define a trend by impression and you decide the exit under the pressure of a loss; define it structurally and you exit by a rule written calmly. See stop loss and confirmation bias, where the absence of a written invalidation is itself the signature.

Three properties make structure a sound base for the rest of the reading:

First, it has no parameters. An indicator such as RSI behaves differently as its period changes, so two settings give contradictory signals on identical data. A swing point is not a setting but a location.

Second, it holds across timeframes. Changing timeframe changes every candle, while major swings stay near where they were. Structure therefore supports multi-timeframe reading where candles do not.

Third, it precedes interpretation. Schools of analysis disagree about what to call structural events and what they imply, yet all of them build on the same highs and lows. Master the base layer and you can evaluate any school instead of memorising its vocabulary — which is exactly why this page stays methodology-neutral: it describes what can be measured and leaves interpretations to their own pages, labelled as interpretations.

Structure is a solid base and not a forecasting tool, and three limits should be clear before building on it.

First, structure lags by nature. A swing point is only known after it has formed and the candles confirming it have closed, so you are describing what happened rather than what will. Expecting an early signal from structure asks it for something it does not offer.

Second, the swing definition is a judgement. There is no agreed candle count, so two readers with different definitions see different structures on the same chart — both internally consistent. That does not invalidate the tool, but it does mean "structure says X" is incomplete without stating the definition used.

Third, the in-between state is real and common. Not every moment is a clear trend or range; much of the time sits in a grey zone with neither consistent progression nor clean boundaries. The correct reading there is to say "undefined" and wait, rather than force a label onto data that does not carry one. Insisting on labelling every moment is confirmation bias wearing analytical clothing.

Order matters, and most reading errors come from starting at step three:

1. Fix your swing definition — how many candles either side, and do not change it between readings.
2. Identify points alternately — high, low, high. Two consecutive highs mean a low was missed.
3. Label each against the previous of its kind — leaving the first of each kind unlabelled.
4. Read the state and extract the invalidation — the last higher low in an uptrend, the last lower high in a downtrend.

Step one is the most neglected: changing the swing definition between readings produces different structures from identical data, making the analysis untestable — and therefore unimprovable.

Once the state is known, the appropriate strategy is known with it: buying dips is coherent inside a range and destructive in a downtrend, while chasing breakouts works in a trend and bleeds in a range. The first question before any signal is not "what is the signal?" but "which state am I in?"