Trends and ranges

The two shapes 

The first is conventionally called a trend, the second a range. Almost anyone recognises them on sight, which is the problem: recognition on sight is not repeatable, and two people reading the same bars routinely produce two different labels. Two things are worth fixing before the definitions arrive. Neither label describes speed, size or steepness, since a trend can be built from small bars over months and a range from violent ones over an afternoon. And both are applied to bars that have already printed, which makes each a description of a completed record.

What a trend is, stated in highs and lows 

The definition rests on two lists rather than on the line the eye draws through them. On any stretch of chart there are points where the price turned down and points where it turned back up. Taken in the order they occurred, they form a list of peaks and a list of troughs. A trend is a stretch in which both lists move in the same direction.

An uptrend is a stretch in which each peak sits above the peak before it and each trough sits above the trough before it. Advancing peaks alone are not enough, and neither are advancing troughs: the ordinary definition requires both lists at once. A downtrend is the same statement with every comparison reversed. The two are strictly symmetrical, and nothing makes one the normal case and the other an exception.

Notice what the definition leaves out. A stretch is an uptrend whether the price climbed in three long bars or in two hundred small ones, and whether it covered a fraction of a percent or a multiple of it. The order of the extremes is what identifies a trend. Steepness does not appear in the definition at all, and that is where the eye and the arithmetic first part company.

One thing the definition leans on and does not supply is a rule for which points count as turning points. The plain reading, a bar standing above the bars immediately either side of it, is enough to classify a stretch at a glance. It is not enough to make two readers agree on every case, because a wider comparison window admits fewer peaks and a narrower one admits more. Identifying turning points precisely is a question in its own right, treated on its own terms elsewhere in this module.

What a range is 

A range is a stretch in which neither list goes anywhere. The peaks stop making progress and cluster around one level, the troughs cluster around another, and the two bound a band the price travels inside. The sequence of extremes is neither ascending nor descending. It is unordered, and that absence of order is the whole of the definition.

Two misreadings are common enough to name. The first is that a range means a quiet market. It does not: the band can be wide, and one move across it can be larger than anything inside a trend on the same chart. Sideways is a statement about the sequence of extremes, never about the size of the bars. The second is that the boundaries of a band are exact levels. Peaks gather around a level rather than stopping at one, and how wide a cluster may be before it stops being a cluster is a judgement, which is where two readers most often disagree.

Consolidation is the word used when a range is described as an interval between two other stretches, which is a larger claim than range makes. Range describes a shape. Consolidation describes the same shape and adds an assertion about its place in a longer sequence, an assertion that cannot honestly be made until the stretch after it has printed. Much writing on charts uses the two words interchangeably. They are not interchangeable.

Reading the sequence off a chart 

Applied to a chart, the definitions become a short procedure with no interpretation in it. The turning points are marked in the order they occurred, the peaks read off as one list and the troughs as another, and each entry compared with the entry before it. Both lists advancing throughout is an uptrend, both declining is a downtrend, neither moving in a single direction is a range.

Run over a stretch whose extremes are not ordered, the same procedure returns the other label.

Nothing in either table is an opinion. Two lists are read in order, each entry is compared with the one before it, and the comparison produces a word. The last two rows of the second block carry the qualification: a single move covers the whole width of the band, so going nowhere refers to where the extremes finished, not to how far the price travelled to finish there.

Why the label is only ever retrospective 

The classification is made from completed turning points, and a turning point is complete only once the bars after it have printed. A peak is not identifiable while the price is still making it. It becomes one when the price has moved away and left it standing. The most recent extreme on any chart is therefore the least certain thing on it, and every label built on it inherits the uncertainty. A label can be withdrawn by bars that have not printed, while every bar already on the chart stays as it was.

Nothing in the first half of that table was wrong. The lists did advance, and the classification described the bars available when it was made. What changed was the length of the record, and the label follows the record. It holds for every classification in this module: a chart label is a statement about the bars to the left of it, and it revises as bars are added to the right.

What the timeframe does to the label 

The same period can carry both labels at once, and that is not a defect in the definitions. A chart is an aggregation, and the aggregation decides which turning points exist. A fall lasting most of a session can sit inside one daily bar's lower wick, where it produces no turning point and enters neither list. On an hourly chart of that same session, the fall contains a dozen peaks and troughs, and they may decline throughout.

So a daily uptrend is routinely built out of hourly stretches, some of them downtrends and many of them ranges, and a daily range opened on a shorter aggregation is not quiet inside: it is a series of complete trends running in both directions between the boundaries of the band. Both readings are correct, because they answer different questions about the same prices.

Two things follow. Trending is not a property of an instrument but of an instrument, an aggregation and a window together, and the window weighs as much as the aggregation: one chart read over two hundred bars and the same chart read over twenty produce different labels as readily as two timeframes do. The second is less comfortable. Switching the timeframe changes the label without any new information arriving. Nothing happened in the market between the two readings. The bars were regrouped.

Where practitioners disagree 

Three parts of this are genuinely unsettled, and a reader will meet all three stated as though they were settled. The first is how many advancing pairs make a trend. Some traditions treat two consecutive higher peaks with two consecutive higher troughs as sufficient, others require three, others count only from the point at which a previous sequence stopped being ordered. No threshold derives from a property of the data, and on identical bars they place the start of a trend in different places and sometimes disagree about whether there is one.

The second is whether both lists have to advance. The definition used above requires it and is the common one. A looser convention accepts advancing peaks alongside troughs that are flat or that overlap the previous trough, on the reasoning that the peaks carry the information. On the same bars the strict convention reports a range where the loose one reports a trend. Neither is a misreading of the chart. They are different definitions producing different words, which argues for stating which definition is in use rather than for settling which is correct.

The third is whether the two categories are exhaustive. A great many stretches fit neither cleanly: peaks advance while troughs decline, or the whole band drifts as it travels. Some practitioners add a third label for that residual and some force every stretch into one of the two. A scheme with no residual category makes its labels look more decisive than the bars underneath them are, and that cost never appears in the label itself.

A fourth observation sits underneath the other three, and it is why reading more charts does not resolve them. Classifications are almost always demonstrated on stretches selected after the outcome was known, so a chart chosen to illustrate an uptrend duly illustrates one. The definitions here are worth learning because they make a description precise and repeatable, and because every later lesson in this module refers back to them. Precision about what has printed is not evidence about what follows it.