Guide · Technical analysis
Fibonacci retracement, explained: the ratios, the mechanism, and the honest limits
The short answer
A Fibonacci retracement marks how far a pullback has undone a prior swing, using horizontal lines at ratios pulled from the Fibonacci sequence. Dividing a term by the next converges to about 0.618, the reciprocal of the golden ratio; two places ahead gives about 0.382, three ahead about 0.236, and the square root of 0.618 gives about 0.786. The widely drawn 50 percent line is not a Fibonacci ratio at all: it is a Dow-theory halfway mark. There is no law forcing markets to honour these levels; the effect is best understood as partly self-fulfilling and strongest at confluence, where a Fibonacci level coincides with an independent support, moving average or trendline.
Fibonacci is the most-drawn and most-misread tool on an Indian retail chart. Almost every explanation lists the levels and stops there, which is exactly the part that teaches nothing. The interesting questions are upstream and downstream of the list: where do 61.8 and 38.2 percent come from, why is a number as clean as 50 percent an impostor, and what is actually happening when price reacts to a line that has no physical claim on it. This guide derives the ratios from the sequence, shows how the tool is anchored to a swing on a Nifty-style move, and is honest about the mechanism, which is confluence and crowd behaviour, not prophecy.
Where the ratios come from: the sequence, then the golden ratio
The Fibonacci sequence starts 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, and continues by a single rule: each term is the sum of the two before it. The retracement ratios are not chosen; they fall out of the relationships between terms as the sequence grows, and they converge to fixed values.
Divide any term by the one immediately after it and the answer settles on roughly 0.618. This is the reciprocal of the golden ratio, written with the Greek letter phi and equal to about 1.618. Skip one term and divide by the one two places ahead and you approach 0.382. Skip two and you approach 0.236. Take the square root of 0.618 and you get about 0.786. Those four numbers, expressed as percentages, are the real Fibonacci retracement levels: 23.6, 38.2, 61.8 and 78.6 percent. The convergence is what matters: it means the ratios are properties of the sequence itself, not of any particular pair of numbers.
Two things follow. First, 61.8 percent is the golden-ratio level, the one direct expression of phi, which is why it is the most watched line on the tool. Second, 50 percent is an impostor in Fibonacci clothing: it comes from Charles Dow's much older observation that a trend tends to give back about half its move, and it was added to the Fibonacci set purely because a halfway mark is useful and sits neatly among the real ratios. Platforms draw it in the same style, so it is routinely, and wrongly, called a Fibonacci level.
| Level | How it is derived | What a pause here tends to say |
|---|---|---|
| 23.6% | Term divided by the term three places ahead (e.g. 34 ÷ 144) | A very shallow pullback; the trend barely paused |
| 38.2% | Term divided by the term two places ahead (e.g. 34 ÷ 89) | A modest pullback in a move with strong momentum |
| 50% | Not Fibonacci: a Dow-theory halfway mark | The midpoint; watched by convention, not by ratio |
| 61.8% | Reciprocal of the golden ratio φ (e.g. 34 ÷ 55) | The classic deep retracement; the most watched line |
| 78.6% | Square root of 0.618 | A deep pullback; near the point that would invalidate the swing |
What a retracement is, and how it is anchored
A market rarely travels in a straight line. It advances, then pulls back part of the way, then resumes or reverses. A retracement is that partial giving-back, and the Fibonacci tool simply measures it as a percentage of the completed swing. The whole exercise hangs on one act: anchoring the tool to the two ends of a swing you can actually see.
For an up-swing you anchor the start of the tool at the swing low and the end at the swing high. The platform then treats that vertical distance as 100 percent and paints horizontal lines back down at 23.6, 38.2, 50, 61.8 and 78.6 percent of it. A pullback that stalls, or turns, as it reaches one of those lines is said to have respected that level. For a down-swing the anchoring is inverted, high to low, and the lines mark how far a bounce has retraced the fall. The judgement is entirely in the anchor: pick a different swing low or high and every level moves, which is why two chartists can draw contradictory Fibonacci levels on the same screen.
Why it can "work", and the honest caveat
Here is the part most Fibonacci pages will not say plainly. There is no established physical or economic law that compels a price to reverse at 61.8 percent of a prior move. The sequence appears in some natural growth patterns, but a stock index is a crowd of humans and machines pricing expectations, not a sunflower head. So when price does react at a Fibonacci level, the honest question is why, and the most defensible answer is uncomfortable for anyone selling certainty.
The effect is, in large part, self-fulfilling. Because a very large number of traders, and the algorithms built on their behaviour, draw the same tool on the same obvious swings, their resting orders, buy limits, stop placements, profit targets, tend to cluster around the same levels. That clustering of real orders is itself capable of producing the pause or bounce the level supposedly predicted. The level does not move the market; the crowd watching the level does. Academic work on the question is consistent with this reading: studies find no single Fibonacci level with strong standalone statistical significance, with retracements distributed fairly continuously rather than spiking at the magic ratios. That is precisely why the professional posture is to treat Fibonacci as context, never as a trigger.
Which leads to the one idea that separates useful Fibonacci from chart astrology: confluence. A Fibonacci level is worth more when it lands on top of an independent reason for price to react, a prior swing high or low, a round number, a moving average that traders watch, or a trendline. When several unrelated groups are watching the same price for different reasons, the cluster of orders is larger and the zone is genuinely stronger. A Fibonacci line sitting alone in empty space, with nothing else near it, is the weakest possible case, and it is exactly the line that price so often ignores.
A worked reading on a Nifty-style swing
Take an illustrative up-swing on the index: a rally from a swing low near 22,000 to a swing high near 24,000, a range of 2,000 points. The retracement levels are just fractions of that range subtracted from the high. Deriving them, rather than reading them off a platform, makes clear how mechanical the tool is, and how loosely its outputs should be held.
| Level | Points given back | Price on the index | How it is computed |
|---|---|---|---|
| 23.6% | 472 | 23,528 | 24,000 minus 0.236 × 2,000 |
| 38.2% | 764 | 23,236 | 24,000 minus 0.382 × 2,000 |
| 50% | 1,000 | 23,000 | 24,000 minus 0.500 × 2,000 (Dow halfway, not Fibonacci) |
| 61.8% | 1,236 | 22,764 | 24,000 minus 0.618 × 2,000 |
| 78.6% | 1,572 | 22,428 | 24,000 minus 0.786 × 2,000 |
Read it as a set of zones to watch, not price targets. A pullback that holds around 23,236, the 38.2 percent line, describes a strong trend that surrendered little; a slide to 22,764, the golden-ratio 61.8 percent line, has given back most of the advance and sits close to the point where the whole up-swing is in question. What the numbers cannot tell you is whether price will honour any of them. That still depends on what else is at each price, an old high, a moving average, a psychological round number, and on the market structure the swing sits inside. The arithmetic is certain; the reaction is not.
Extensions: projecting beyond the swing
The same sequence supplies levels above 100 percent, used not to measure a pullback but to project a continuation. These are Fibonacci extensions. The two most cited are 127.2 percent, which is the square root of the golden ratio 1.618, and 161.8 percent, the golden ratio itself. Drawn from a swing and its retracement, they mark rough zones a trend might reach if it carries on, so traders use them as reference targets.
The caveat is identical, and worth stating in the same breath as the levels themselves: an extension is a projection, not a promise. It inherits every limitation of retracements, no physical claim on price, a largely self-fulfilling and continuous character, and it is most useful where an extension level coincides with an independent structural level such as a weekly high or a measured-move target from a chart pattern. Treat 161.8 percent as a place to pay attention if price gets there, never as a forecast that it will.
| Aspect | Retracement | Extension |
|---|---|---|
| Measures | A pullback inside a completed swing | A projection beyond the swing |
| Typical levels | 23.6, 38.2, 50, 61.8, 78.6 percent | 127.2, 161.8, 261.8 percent |
| Range | Between 0 and 100 percent of the swing | Above 100 percent of the swing |
| Used as | Zones where a pullback may pause or turn | Rough target zones for a continuation |
| Same caveat | Context, not prediction; strongest at confluence | Context, not prediction; strongest at confluence |
How the tool is misused, and how it fails
Most of the damage done with Fibonacci comes from treating an aid to judgement as a source of it. Two failure modes account for nearly all of it, and both are baked into how the tool looks rather than how markets behave.
The first is the five-line trap. Draw 23.6, 38.2, 50, 61.8 and 78.6 percent on any swing and you have blanketed the pullback zone so thoroughly that price is almost guaranteed to touch one of them. After the fact, you can always point to the level it respected and feel the tool worked. Before the fact, five candidate lines give you no actionable answer about which, if any, matters. This is the essence of a post-hoc explanation dressed as a prediction.
The second is anchor subjectivity. Because every level is a fraction of the swing you chose, the entire grid shifts when you pick a different swing low or high. Two competent chartists can produce contradictory Fibonacci maps of the same chart, each internally consistent, simply by disagreeing about which swing to measure. There is no objective anchor, which means Fibonacci can be quietly fitted to almost any narrative.
Where Fibonacci breaks. Price does not respect lines, it respects orders and information. A strong trend can blow through 61.8 percent without pausing; a news shock can gap past every level at once; in a choppy, directionless market the levels light up at random and mean nothing. And the levels are zones, not exact prices, so a reaction "at" 61.8 percent is often several points early or late. Any use of Fibonacci that ignores the market structure around it is decoration, not analysis.
Where Fibonacci fits, and what it is worth
Placed correctly, Fibonacci is a genuinely useful context tool. It gives a disciplined vocabulary for how deep a pullback is, it flags prices that a large part of the market is likely watching, and, used at confluence, it helps grade a zone you already identified from structure. Those are real contributions to reading a chart. None of them is a licence to act on a level in isolation.
The honest framing is that Fibonacci sits on top of analysis, never underneath it. The load-bearing work is upstream: identifying the swing that matters, reading the trend and the structure it lives in, and locating the prior levels that give a Fibonacci line something to agree with. Get that right and the ratios sharpen the picture; get it wrong and no ratio can rescue it. That upstream judgement, structure first and indicators as confirmation, is exactly what the method we teach is built around. Fibonacci is a lens, not a map, and a lens is only as good as what you point it at.
Common Questions
Frequently Asked Questions
Where do the Fibonacci retracement ratios actually come from?
+They are ratios between terms of the Fibonacci sequence, where each term is the sum of the previous two. Divide a term by the next one and the result converges to about 0.618, the reciprocal of the golden ratio 1.618. Divide by the term two places ahead and you get about 0.382, three places ahead about 0.236. The square root of 0.618 gives about 0.786. Those numbers, 23.6, 38.2, 61.8 and 78.6 percent, are the retracement levels.
Is 50 percent a Fibonacci ratio?
+No. The 50 percent level is not derived from the Fibonacci sequence at all. It is a Dow-theory idea, the old observation that a move often gives back roughly half of its ground, and it was bolted onto the Fibonacci tool because it sits conveniently in the middle. Charting platforms draw it alongside the true ratios, so many traders assume it is one. It is not. The genuine golden-ratio level is 61.8 percent.
What is a Fibonacci retracement and how is it drawn?
+A retracement measures how far a pullback has undone a prior swing. You anchor the tool to the two ends of a completed swing, the low and the high of an up-move, and the platform draws horizontal lines at 23.6, 38.2, 50, 61.8 and 78.6 percent of that range. A pullback that pauses or turns near one of those lines is said to have respected that level. The lines are zones, not exact prices.
Why do Fibonacci levels seem to work?
+There is no proven physical reason markets must obey these ratios. The most defensible explanation is that the effect is partly self-fulfilling: because a large number of traders and algorithms watch the same levels, resting orders cluster there, and that clustering can create the very reaction the level predicted. Studies find no single level with strong standalone statistical significance, which is why practitioners treat Fibonacci as a context tool rather than a signal.
What is Fibonacci confluence?
+Confluence is when a Fibonacci level lands on top of another independent reason for price to react: a prior swing high or low, a round number, a moving average, or a trendline. A 61.8 percent retracement that coincides with an old support shelf is a stronger zone than either factor alone, because several groups of traders are watching the same price for different reasons. A Fibonacci level with nothing else near it is the weakest case.
Which Fibonacci level is the most important?
+The 61.8 percent level is the one most watched, because it is the direct golden-ratio level, the reciprocal of 1.618. The 38.2 percent level is the next most cited. A shallow pullback to 38.2 percent suggests a strong trend that gave back little; a deep pullback to 61.8 or 78.6 percent has undone most of the move and sits closer to invalidating the swing entirely. None of this makes any level a place to act on its own.
What is the difference between a Fibonacci retracement and an extension?
+A retracement measures a pullback inside a completed swing and its levels fall between 0 and 100 percent. An extension projects beyond the swing, using levels such as 127.2 percent, the square root of 1.618, and 161.8 percent, the golden ratio itself, to mark where a continuation might reach. Extensions are used as rough target zones, and they carry exactly the same caveat as retracements: they are reference points, not predictions.
Can I trade using Fibonacci levels alone?
+Using them in isolation is the classic misuse. Five levels drawn on any swing almost guarantee that price touches one, which makes it easy to explain any move after the fact and hard to act on any level before the fact. Price routinely slices straight through a lone Fibonacci line. The tool earns its place only as one input among structure, prior levels and trend, never as a standalone entry rule.
Does Fibonacci retracement work on the Nifty?
+The Nifty 50 is a liquid, widely followed index, so it is exactly the kind of instrument where the self-fulfilling mechanism is plausible: many participants draw the same retracements on the same swings. That makes Fibonacci a reasonable context tool on the index, strongest where a level coincides with an obvious prior high or low. It does not make the levels predictive, and it is no substitute for reading the market structure underneath.
Where the facts come from
Sources
- Derivation of the ratios. The 23.6, 38.2, 61.8 and 78.6 percent levels are obtained by dividing a term of the Fibonacci sequence by the term one, two or three places ahead (0.618, 0.382, 0.236) and by taking the square root of 0.618 (0.786); 0.618 is the reciprocal of the golden ratio 1.618. chartschool.stockcharts.com
- 50 percent is not a Fibonacci ratio. The 50 percent line is included by convention from Dow theory, Charles Dow's observation that a move tends to give back about half its ground, not from the Fibonacci sequence. financestrategists.com
- The self-fulfilling and confluence reading. There is no established mathematical reason markets must respect Fibonacci levels; the effect is widely attributed to a self-fulfilling clustering of orders from the many traders watching the same levels, and studies find no single level with strong standalone significance, so the levels are used in confluence with other tools. digitalcommons.macalester.edu
Educational note. This guide explains a technical-analysis tool and the mathematics behind it. It is not a recommendation to trade or invest, is not a signal or a price forecast for any index or security, and is not investment advice. Bharath Shiksha is an educational publisher, not a SEBI-registered investment adviser or research analyst. Chart examples are illustrative.