6 / 7 · 8 min
Patterns: some have a mechanism, others only a name
Behind triangles and flags stands a measurable thing — volatility compressing before it expands. Behind most other patterns stands only tradition and the eye's talent for seeing faces in clouds.
A pattern is a name given to a recognisable shape of quiet. Its usefulness depends entirely on whether the name describes a real mechanism or only an appearance. One question separates them: what exactly must the participants be doing for this picture to form? If there is an answer and it can be tested, the pattern works. If the answer sounds like "that is how it has always been", you are holding a name with nothing inside.
The mechanism that genuinely exists
Volatility travels in waves: wide candles are followed by narrow ones, narrow ones by wide again. This is not mysticism or a property of charts — it is how a market behaves when participants alternate between disagreeing and agreeing. Once agreement is reached the range of candles collapses; once new information arrives it explodes. Every working pattern — triangle, flag, rectangle — is a different outward appearance of the same compression. Compression is what can be measured, not the shape.
The ratio of recent range to the usual one. A value near one means the market is behaving as it always does. Below 0.6 means candles have become noticeably narrower than usual: compression is under way. Example: a coin travels between 1,180 and 1,240, ATR over five candles has fallen to 18 against a usual 41 — a ratio of 0.44. Now the important part: compression says expansion is approaching and does NOT say in which direction. Anyone claiming otherwise is selling you a direction that is not in the data.
What each pattern describes
| Pattern | What is actually happening | What it needs to mean anything |
|---|---|---|
| flag, pennant | a pause in a trend: early entries take profit, no new sellers | falling volume inside the pause |
| triangle | compression: both sides converge on one price | narrowing that is real in ATR, not in your lines |
| rectangle | a range: the boundaries hold orders | visible boundaries without your drawings |
| head and shoulders | a failed new high plus a structure break | nothing beyond what the structure already says |
Head and shoulders: an honest reading
On closer inspection the most famous reversal pattern turns out to be a retelling of the first lesson. The left shoulder and the head are rising highs. The right shoulder is a high that did NOT exceed the previous one. Breaking the neckline is a move below the last significant low. In other words, this is an ordinary structure break given three words instead of two. The name adds not a single new fact. But there is one honest caveat: everybody knows this pattern, so real stops and limit orders do pile up along the neckline. It works not because it predicts but because people watch it.
A target measured by the pattern's height
The rule "project the height of the pattern from the breakout point" is repeated so often that it feels like a law. It has no mechanism: the height of a consolidation is the range of an argument over the past few days and bears no relation to where price will be carried afterwards. The reason the rule survives is simpler: on average a pattern's height is proportional to the instrument's volatility, and after a breakout price on average travels a distance proportional to that same volatility. There is a correlation, not a cause. Compute the target from ATR directly and you get the same number without the mythology.
Seeing a pattern where there is none
The human eye finds faces in clouds and works the same way on a chart. Take any fifty candles and try to draw a triangle — you will succeed, and in several ways at once. That is the real problem with patterns: not that they fail to work, but that they can be found everywhere. There is one defence and it is numeric: before naming a shape, measure the compression. If ATR is not falling there is no triangle, however convincingly the lines converge.
What to do with a compression breakout
The direction cannot be guessed, but you can prepare for both. The practical approach: mark both boundaries of the compression and decide in advance what you do on an exit upward and what on an exit downward. The stop then goes beyond the opposite boundary rather than behind the nearest candle — inside a compression price thrashes about, and a tight stop there is taken out by noise. And remember the most common development: the first exit is often false, and the real move begins after it in the opposite direction. That is why entering on confirmation costs more but survives this trap.
Compression is visible in numbers, not by eye: the table has a NATR column — an instrument's ordinary range relative to its price. Falling NATR alongside a narrow daily travel is exactly the consolidation that a pattern merely draws. On the chart a consolidation is conveniently marked with a rectangle: it stays put while you switch scales and shows immediately whether the boundaries really narrowed.
The chart and its drawing toolsFind ten compressions by number, not by eye
Sort the table by NATR ascending and take the ten least volatile coins among those with decent turnover. Open each and see what the chart looks like. Then mark the boundaries of the quiet with a rectangle and come back a few days later. Count two numbers: how many compressions ended in an exit, and how many times the first exit turned out to be false. The second number usually surprises more than the first — and it is the one that decides whether you should trade breakouts at all.
The triangle's lines converge perfectly, but ATR over the last five candles has not changed. What does that mean?
That the triangle was drawn by you and not by the market. Converging lines can be fitted to almost any stretch of chart: you pick the points, and suitable points always exist. Real compression means the range of candles has genuinely shrunk, and that shows up in a number, not in a drawing. If ATR is flat there is no mechanism, and there is no reason to expect expansion from such a pattern.
If head and shoulders is just a structure break, why know the pattern at all?
For exactly one reason: everybody else knows it. The neckline enters other people's trading plans, and stops and limit orders genuinely accumulate along it — that is, it becomes a level in the same sense that last week's high is one. The pattern is worth knowing not for prediction but for understanding where other people's money will be sitting. This, by the way, is a general rule: any widely known marking is valuable first of all because many people see it.