Retracement: The Move That Does Not End a Trend
A retracement is a move against the prevailing trend that ends without reversing it, after which the trend resumes. It is distinguished from a reversal only by what happens afterwards, which makes depth and context the only things available to judge it by while it is happening.
How it works
Trends do not move in straight lines. An uptrend advances, gives some of it back, advances again. The giving-back part is the retracement.
The definition contains its own problem. A retracement is a counter-trend move that does not end the trend — which can only be established once the trend has resumed. At the moment it is happening, the thing that defines it has not happened yet.
So everything on this page is about handling that ambiguity, not about resolving it. There is no measurement that separates a retracement from the beginning of a reversal in real time, and any material claiming otherwise is describing hindsight.
Set the two side by side and the first half is the same chart. Price rises, then falls. In one case it stops falling and rises again; in the other it does not. The information that separates them arrives after the decision point.
Depth is the one measurable property
Take the leg that just happened — low to high — and measure how far price has come back. A third of the way. Half. Two-thirds. That percentage is the retracement depth, and it is the only property that can be stated precisely while the move is in progress.
What depth tells you is about the trend, not about the future. Shallow retracements are characteristic of strong trends: buyers step in early and price never gives much back. Deep ones are characteristic of trends losing conviction, or of trends about to end — and there is no threshold that separates those two cases.
Fibonacci retracement levels are this measurement with fixed percentages attached — the familiar 38.2%, 50% and 61.8%. The Fibonacci page covers where they come from.
What they are: a consistent way to describe depth. Saying “price retraced to the 61.8%” is more precise than “price came back quite a long way,” and precision is worth having.
What they are not: a reason for price to stop. The levels are drawn by the observer after choosing which swing to measure from, and a different swing choice produces different levels on the same chart. Anything that moves when you change your mind about the starting point is a description, not a cause.
The shallow case is the one worth noticing. A trend that keeps giving back only a small fraction before advancing again is behaving differently from one giving back most of each leg, and that difference is visible without predicting anything.
In practice: what it costs to participate
The orthodox check is volume falling through the retracement — the idea being that a genuine pause has less participation than the trend leg itself. It is a reasonable lean and it is not decisive: plenty of reversals begin quietly and plenty of continuations begin on heavy volume.
Each attempt pays the spread and commission on a round trip. A trend that offers a retracement every few days offers a round trip every few days, and the ones that turn out to be reversals cost the stop as well as the fee.
Timeframe changes the word entirely. A move that is a violent reversal on a five-minute chart is a single red candle on the daily and a rounding error on the weekly. The same price action is a retracement or a reversal depending on which chart you are describing it from — which is worth saying plainly, because it means the disagreement between two traders is often about timeframe rather than about the market.
There is nothing in the order book that marks where a retracement ends. Resting orders sit where people put them, and where people put them is largely at the obvious levels — which produces some genuine reaction at those levels and no obligation whatsoever.
What a retracement is not
It is not a discount. Price being lower than it was is not evidence it is cheap. The trend that made it higher is the same trend that may have ended.
It is not defined by a percentage. No depth makes a move a retracement rather than a reversal. A 20% giveback that keeps going is a reversal; an 80% giveback that resumes is a retracement.
It is not the same as a correction. In common usage a correction is a larger, index-level decline with its own conventional thresholds. A retracement is a structural description that applies at any scale.
And it is not a pattern. There is no shape to recognize. It is a description of where price is relative to the leg before it.
How this connects
Market structure is how it gets labeled. A pullback that holds above the prior higher low keeps the sequence of each higher high intact, while a break below that low is a break of structure and often the first sign of a change of character.
Levels do the practical work. Traders commonly look for a retracement to end at support and resistance, often where role reversal turns a broken ceiling into a floor, or at a moving average.
The ATR puts depth in context. A move measured in multiples of average range is comparable across instruments in a way a raw percentage is not.
The stop is where it becomes a reversal. A stop loss beyond the last swing high or low declares the point at which the idea is wrong, and position sizing follows from that distance.
And it is the setting for adding to a trend. To scale into a position is to buy retracements deliberately, which is why the risk per trade has to account for every addition.
When it fails
In a range the concept dissolves. Price alternates between boundaries, so every leg retraces the previous one, and there is no trend for anything to be a retracement of. The vocabulary keeps working while the meaning has gone.
The second failure is treating depth as a signal. Buying at 61.8% because it is 61.8% is buying a number chosen by an observer, and the level moves if the observer picks a different swing.
A third is averaging into it. A retracement that becomes a reversal punishes exactly the behavior this frame encourages — adding at progressively worse prices while the story stays intact.
A fourth is measuring from the wrong leg. Which swing counts as “the” prior move is a judgment, and on any chart there are several defensible choices producing several sets of levels. Picking the one that puts a level where price already stopped is drawing the target around the arrow.
And a fifth is expecting symmetry. Retracements in an uptrend and in a downtrend do not behave the same way — declines are typically faster and deeper — so a framework calibrated on one direction misreads the other.
The original data
The corpus of 24,971 videos measured for this site covers retracement almost entirely through Fibonacci content rather than as the structural measurement it is — which is how a way of describing depth became, for most people, a set of lines that are supposed to do something.
The measurable content here is depth and cost. Depth is computable on any chart in seconds and tells you something real about how the trend has been behaving — the S&P 500’s 56.8% fall from 1565.15 in October 2007 to 676.53 in March 2009 still reads as a retracement on a chart running from 1950 to 2026, because the index closed above that old peak on 28 March 2013. Cost is the spread and commission on every attempt, which is what decides how many retracements a trend has to offer before participating in all of them stops being worth it. What is not measurable — which retracement will resume and which will not — is the part every article about this promises and none of them delivers.
Related
Pullback is the same event framed as a trade rather than a measurement. Fibonacci is where the depth percentages come from. And trend following is the approach that depends on retracements resuming.
I do not treat every retracement as an opportunity. They are only opportunities in trends that continue, and there is no reliable way to know in advance which ones those are, so buying every dip is buying weakness and calling it discipline.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.