Fibonacci retracements and extensions

After a stock rallies, it almost never goes straight up — it gives some of the move back before (maybe) continuing. The natural question every trader asks is: how far back will it fall before it turns? Fibonacci retracements are the most popular attempt to answer that, and you will see their levels — 38.2%, 50%, 61.8% — drawn on charts everywhere. This module explains where those numbers come from, how people use them, and then walks straight into the argument that will not go away: are these levels real, or do they only "work" because everyone believes in them?

The numbers come from a sequence found by Leonardo of Pisa, nicknamed Fibonacci, in the 1200s: 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, and so on, where each number is the sum of the two before it. As the sequence grows, the ratio between neighbouring numbers settles toward a constant — roughly 1.618, and its inverse roughly 0.618. This is the golden ratio, and it does genuinely appear in some natural growth patterns, from sunflower seed spirals to shell shapes.

From these ratios, traders derive the retracement percentages: 61.8% (the ratio itself), 38.2% (its square), 23.6% (the next power), and 50% — which is not actually a Fibonacci number at all, but is included by convention because a halfway pullback is psychologically natural.

The leap — and it is a genuine leap — is the belief that because this ratio appears in seashells, it should also govern how far a stock pulls back. There is no accepted mechanism connecting the two. That gap between "the ratio is real in nature" and "the ratio rules markets" is the whole debate, and we will not paper over it.

A Fibonacci retracement is drawn between two points: the low and the high of a swing you care about. Software then places horizontal lines at the Fibonacci percentages of that range. Each line marks a price where the pullback might pause before the trend resumes.

Here is a clean uptrend from 100 to 150, with the retracement levels drawn across it. Watch where the pullback finds its footing.

The band between the 38.2% and 61.8% levels — sometimes called the "golden pocket" — is where retracement-watchers expect the strongest reactions. A shallow pullback to 23.6% suggests a very strong trend; a deep one past 61.8% suggests the trend may be in trouble.

The mirror idea is the Fibonacci extension: instead of measuring how far price pulls back, it projects how far price might run beyond the old high, at ratios like 127.2%, 161.8% and 261.8% of the original swing. Traders use extensions to set rough profit targets.

Notice, in the chart above, that the pullback paused near 50% — but also that 50% is not even a true Fibonacci number. If a non-Fibonacci level works as often as a Fibonacci one, that is already a clue about what is really going on.

There are two honest explanations for why a retracement level sometimes holds, and neither requires magic. The first is self-fulfilling behaviour: so many traders draw the same levels from the same obvious swing that a crowd of buy orders clusters at, say, 61.8%. Price pauses there because people made it pause — the level works because it is watched, not because the number has power. The second is confluence: the level happens to land on top of prior support, a round number, or a moving average, and it is that stack of ordinary reasons, not the Fibonacci ratio, doing the work.

Fibonacci levels cannot tell you which swing to measure. Every chart has many highs and lows, and different choices of start and end point produce different levels. This freedom is the method's quiet weakness: with enough candidate swings, some level will always sit near wherever price happened to turn, and it is tempting to draw the retracement after the fact so it appears to have predicted the turn. That is curve-fitting, not forecasting.

They cannot tell you the level will hold. A retracement marks a place where a reaction is a little more likely — nothing more. A level nudges the odds of a pause; it never promises one. In a powerful trend or a fast decline, price treats these lines as if they were not there.

And they cannot deliver the precision they appear to promise. A line drawn to one decimal place looks scientific, but the "reaction" often comes several rupees above or below it, so practitioners really watch a zone, not a price. The apparent exactness is mostly an illusion of the software. The clean number hides a fuzzy reality; treat the level as a rough area of interest, never a magic price.

The first way people are fooled is by hindsight drawing. It is trivial to place a Fibonacci retracement so that a level lands exactly where price already turned — you simply choose the swing that makes it fit. The chart then looks like proof the levels work, when all it proves is that you drew them afterwards. The honest test is a level chosen before the pullback, from the obvious swing, and then left alone.

The second is treating the levels as certainties instead of soft zones. A trader buys "the 61.8%" with no stop because the textbook said it is strong support, and rides a position straight down when this particular pullback keeps falling. The level was a place to watch, not a place to trust blindly.

The third is mysticism — the belief that because the golden ratio appears in nature, it must secretly rule the market. This feels profound and explains nothing. The far more likely truth is mundane and useful: these levels are shared reference points that a watching crowd sometimes honours, strongest when they coincide with other evidence, and worthless in isolation. Used that way — as one input in a confluence, with a stop — they are a reasonable tool. Believed as magic numbers, they become a story that will eventually cost you.