No tool in technical analysis attracts more mystique than Fibonacci retracements — and none needs demystifying more. Strip away the golden-ratio romance and what remains is a genuinely practical instrument: a systematic way of mapping where a pullback sits relative to the move that preceded it, so that traders across the market are watching the same shelves at the same time.
This guide covers where the levels come from, how to draw them properly (the step most people get wrong), what each zone conventionally means, how professionals combine them with other evidence, and the case for and against the whole exercise, argued both ways.
Where the numbers come from
The Fibonacci sequence — 1, 1, 2, 3, 5, 8, 13, 21, 34, 55… — builds each term by adding the previous two. As the sequence grows, the ratio between consecutive terms converges on 1.618, the number the Greeks called the golden ratio. Its reciprocal is 0.618. Take a term two places back and the ratio tends to 0.382; three places back, 0.236.
From these come the standard retracement levels: 23.6%, 38.2%, 50%, 61.8% and 78.6% (the square root of 0.618). The 50% level is not a Fibonacci number at all — it earned its place from the older Dow-theory observation that healthy trends often give back about half their gains before resuming. Traders keep it because it works well enough as a reference, which tells you something important about this tool: the levels are conventions with history, not laws of nature.
Why should stock prices care about ratios found in sunflower spirals? The uncomfortable-but-useful answer: partly they don't, and partly they do because everyone is looking at them. When lakhs of participants draw the same lines from the same swing points, orders cluster around those prices — buyers waiting at "the 61.8", stop-losses tucked beneath it. The levels gain a measure of self-fulfilling relevance independent of any cosmic significance. A trader does not need to believe in golden ratios to respect where other people's orders sit.
Drawing the retracement correctly
A retracement is always measured on a completed (or provisionally completed) swing — one clear directional move from a swing low to a swing high, or the reverse.
1. Identify the impulse. In an uptrend, find the meaningful swing low where the advance began and the swing high where it paused. Not every wiggle qualifies; you want the move a person glancing at the chart would identify as the recent leg. 2. Anchor low to high (for an up-move). The tool then divides that vertical distance into the standard percentages: a 38.2% retracement means price has given back 38.2% of the leg. 3. For down-moves, anchor high to low. The levels then mark how much of the decline a bounce has recovered — the same arithmetic mirrored.
The classic errors are anchoring errors. Using intraday wicks versus closing extremes changes every level, so pick one convention and stay with it (many practitioners use the actual high/low wicks on daily charts). Anchoring a minor squiggle instead of the dominant swing produces levels nobody else is watching — which defeats the crowd-coordination logic that gives the tool its force. And redrawing anchors every few days until the lines "fit" is not analysis; it is doodling with confidence.
Timeframe matters too. A retracement of a two-year advance marks zones that can hold for months; a retracement of last week's pop is intraday furniture. When levels from different timeframes land on the same price, pay attention — that is confluence, and it is the heart of serious Fibonacci use.
Reading the zones
Convention assigns each band a personality. Treat these as base rates and tendencies, never guarantees:
- 23.6% — the token dip. Trends powerful enough to pause only here are usually being chased by under-invested participants. Frequent in momentum leaders; often too shallow to offer a comfortable entry with definable risk.
- 38.2% — the strong-trend pullback. The classic "healthy correction" in a robust advance. Many continuation setups — flags, three-week pullbacks to a rising 20-day average — bottom out in this vicinity.
- 50% — the psychological midpoint. Half the move surrendered. Enough fear to shake out weak hands, not enough to break the structure. A vast amount of practical support/resistance work happens around halves, Fibonacci or not.
- 61.8% — the golden zone and the line of debate. The last conventional station where a pullback is still a pullback. Bulls defend it loudly; that is exactly why its failure is informative — a decisive close through 61.8% converts many holders' theses from "buying opportunity" to "something is wrong".
- 78.6% — the deep test. Price has taken back almost everything. Occasionally a violent shakeout ends here and the trend resurrects (the "deep retracement, strong hand-off" pattern), but the sober base rate: most moves that give back this much were not resuming trends but topping structures.
Beyond 100% lies extension territory — 127.2%, 161.8% — used for projecting targets rather than retracements, a topic of its own.
Confluence: the actual edge
Professionals rarely trade a Fibonacci level naked. The working method is confluence — waiting for a retracement level to coincide with independent evidence:
- Structural support or resistance. A 61.8% retracement landing exactly on a prior breakout shelf or a multi-month base top is two maps agreeing. TaurEye's screener computes daily, weekly and monthly support/resistance distances from swing pivots — when a Fibonacci zone and a pivot-based level overlap, the shelf is real in both frameworks.
- Moving averages. Pullbacks in strong trends habitually find the rising 50-DMA; when the 50-DMA passes through the 38.2–50% band of the recent swing, the zone thickens.
- Round numbers. Prices like 500, 1000, 2500 attract orders in India as everywhere. A 50% retracement at ₹998 is stronger furniture than one at ₹1,013.
- Volume signatures. A pullback drifting into the zone on shrinking volume, then reversing on an expansion day, is the tape agreeing with the map. A collapse through the zone on heavy volume is the tape overruling it — believe the tape.
- Momentum resets. RSI cooling from the 70s to the 40–50 region while price sits in the 38–50% band is the oscillator's way of saying the excess has been digested.
The general rule: one level is a line, two agreeing levels are a zone, three are a plan. The more independent frameworks point at the same price, the less your outcome depends on any single tool being "right".
A worked example
Imagine a capital-goods stock that ran from ₹640 to ₹940 over eleven weeks — a 300-point leg. It then stalls and begins drifting down on fading volume. The retracement map reads: 23.6% at ₹869, 38.2% at ₹825, 50% at ₹790, 61.8% at ₹755, 78.6% at ₹704.
Price slides for three weeks and stabilises around ₹792–800. Now assemble the context: the 50% level (₹790) sits there; so does the top of the March consolidation (₹785–795), a natural support shelf; the rising 50-DMA has climbed to ₹788; and the round figure ₹800 hovers overhead. Four frameworks, one zone. Daily volume during the decline halved; RSI has cooled from 78 to 46.
Nothing about this guarantees a resumption. What the confluence does is define a high-information location: if buyers are going to defend the trend, this is where their footprints will show (reversal bars, expansion volume, a reclaim of ₹800). And if instead the zone breaks decisively, the map has told you something equally valuable — the pullback has graduated into something larger, with 61.8% at ₹755 as the next reference and the thesis on notice. Either way the trader is responding to evidence at pre-identified prices rather than improvising in the noise.
Fibonacci in downtrends and for exits
The tool mirrors cleanly. In a decline, retracements mark where bounces tend to exhaust: bear-market rallies famously die in the 38.2–61.8% recovery band of the prior fall, which is why "dead-cat bounce into the golden zone" is a staple of short-side playbooks. For position management, some traders use retracement logic in reverse — treating a give-back of more than 61.8% of their open profit's underlying swing as the objective signal that the move they were riding has structurally changed.
Retracements versus "buying the dip"
It is worth naming the difference between this framework and the reflex it superficially resembles. "Buying the dip" as commonly practised has no definition of dip, no invalidation price and no answer to "what if it keeps falling?" — it is averaging down wearing a strategy's clothes. Retracement analysis, done properly, supplies all three: the dip is measured against a specific swing; the zones define where interest is warranted and where the structure fails (a decisive loss of the 61.8% region); and position size is set against the distance to that failure point. Same instinct — trends pull back and resume — but one version is a plan with an exit, and the other is a hope with a cost basis. The lines on the chart matter less than the discipline they scaffold.
The case against — and what survives it
Rigorous studies of Fibonacci levels struggle to show that 38.2% or 61.8% attract reversals more than nearby arbitrary percentages once you account for how often prices pause anywhere. Confirmation bias does heavy lifting: the level that "worked" is remembered, the three that sliced through are forgotten. Anchor choice is subjective enough that two careful analysts draw different maps of the same chart.
All of that is true — and yet the practical defence stands on two legs that do not require mysticism. First, coordination: enough participants watch these exact levels that order flow really does cluster near them, especially on widely-followed index and large-cap charts. Second, discipline: the retracement framework forces a trader to pre-define locations, invalidation points and risk before the emotional moment arrives. A map that is only approximately right but is drawn in advance beats improvisation that is occasionally brilliant. Use the levels as scaffolding for planning, demand confluence and confirmation before acting, and the golden ratio can stay in the sunflowers where it belongs.
Extensions: projecting beyond the high
Retracements answer "how deep might the pullback go?" Their sibling — extensions — answer the forward question: "if the trend resumes, how far might it travel?" The standard extension levels are 127.2%, 161.8% and 261.8% of the original swing, projected beyond its endpoint.
Return to the ₹640→₹940 example. If the pullback holds at ₹790 and the stock reclaims its high, extension targets sit at roughly ₹1,021 (127.2% of the 300-point leg from the ₹790 pivot's perspective differs by method; the simplest projection adds 0.272 × 300 to the old high) and ₹1,125 (161.8%). Methods vary — some measure from the retracement low, some from the original low — which is another reminder that these are planning conventions, not physics. Their genuine value is behavioural: an extension target chosen before the breakout gives a trend follower a pre-committed zone to lighten up in, taming the twin demons of selling brilliance too early and holding euphoria too long.
Extensions also mark where measured-move symmetry lives: the tendency of a second leg to approximate the length of the first (a 100% extension). When a 100% measured move, a 127.2% extension of a smaller swing, and a prior all-time-high shelf cluster together, you have target confluence — the same multi-framework logic as entry confluence, pointed forward.
Fibonacci on the index: why NIFTY levels get so much airtime
Watch any budget-day or results-season broadcast and you will hear index retracement levels quoted with striking specificity. There is a structural reason the tool is more meaningful on NIFTY and BANKNIFTY than on an individual small cap: participation density. Index derivatives are among the most liquid instruments in the country; tens of thousands of participants — institutional desks, prop firms, option writers — mark the same swings on the same charts. Their collective orders around the 38.2% or 61.8% of a well-defined index swing create genuine liquidity shelves, visible in how often intraday moves stall and rotate near broadcast levels.
Two practical consequences. First, on indices, prefer the most obvious swing anchors for the simple reason that obviousness is what recruits the crowd — the pandemic low, the recent all-time high, the correction extremes every analyst cites. Second, on thinly-followed stocks, invert the humility: your beautifully drawn levels may be watched by nobody, so demand stronger non-Fibonacci confluence (structure, volume, averages) before trusting them.
Quick answers to common doubts
Do professionals really use this, or is it retail folklore? Both. Systematic funds generally do not encode golden ratios; discretionary traders, prop desks and a large share of technical practitioners worldwide keep retracement grids on their charts — if only because everyone else does. The tool's institutional footprint is strongest in FX and index futures, weakest in illiquid single names.
Which timeframe's levels win when they conflict? The higher timeframe, as a rule of thumb. A weekly 38.2% zone overrules a daily 61.8% in most practitioners' hierarchies. When a daily bounce fights a weekly ceiling, the weekly usually collects.
Should levels be drawn on closing prices or wicks? Consistency beats correctness — there is no "right" answer, only the discipline of one convention. Wick-to-wick is the more common choice on daily charts because extremes are where stops actually lived.
How much tolerance should I give a level? Scale it to volatility: roughly a quarter to half of the stock's daily ATR on either side is a workable buffer. A ₹5 miss on a stock that travels ₹40 a day is a direct hit; the same miss on a sleepy ₹200 stock is a genuine miss.
Can I screen for stocks near retracement levels? Not directly by ratio in most screeners — but you can approximate the situation: stocks in uptrends (above the 200-DMA) trading within a few percent of pivot-based support, with RSI cooled to neutral, describes "pullback into potential demand" in screenable terms. The Fibonacci grid then becomes your chart-level second opinion.
Common mistakes
- Anchoring trivia. Levels drawn on minor swings coordinate with nobody.
- Trading the touch. A level is a place to watch for evidence, not a buy button. Wait for the reversal signature.
- Ignoring the break. A decisive close through your zone is information — the market disagreeing with your map. Update the map.
- Precision worship. These are zones, not laser lines. Give levels a buffer proportional to the stock's ATR.
- Fibonacci everything. Retracements on 15-minute charts of illiquid small caps measure noise with elegant arithmetic.
A closing word on the mystique
Fibonacci retracements convert the vague question "has this pulled back enough?" into a structured one: "how much of the driving swing has been surrendered, and does that shelf coincide with independent support?" That reframing — from prophecy to cartography — is the entire, sufficient case for the tool. Draw the dominant swing, respect confluence, demand confirmation, and let the mystique remain a marketing story.
Shared purely to explain a charting technique — it is not advice, and no level drawn from past prices guarantees anything about future ones. Before acting on any market view, talk it through with a SEBI-registered adviser.