How to Draw Fibonacci Retracement, and What the Levels Can and Cannot Tell You
Fibonacci retracement is one of the easiest tools to draw and one of the easiest to draw in a way that tells you nothing. The mechanics take thirty seconds. The judgement is in choosing which swing to measure, and in being honest about what the resulting levels are.
The short answer
Fibonacci retracement divides a completed price swing into proportions - most commonly 23.6%, 38.2%, 50%, 61.8% and 78.6% - and marks those as horizontal levels where a pullback might pause. In an upswing you drag from the swing low to the swing high; in a downswing, from the swing high to the swing low. The levels are not magnetic. They become worth watching when they line up with something else, such as a prior support or resistance zone, a trendline or a round number.
What the tool actually does
Mechanically, Fibonacci retracement is nothing more than proportional division. You pick a completed move - a swing from a low to a high, or a high to a low - and the tool draws horizontal lines at fixed fractions of that distance.
The fractions come from ratios associated with the Fibonacci sequence, which is where the name and a lot of the mystique originate. You do not need the mathematics to use the tool. What matters is that it gives you a consistent way to ask: "if price pulls back from this move, how far has it retraced?"
The common levels
23.6% - a shallow pullback. In strong trends, retracements often stay this shallow.
38.2% - a moderate pullback.
50% - not a Fibonacci ratio at all, but included by almost every charting platform because halfway is a natural reference point.
61.8% - the level most traders watch. The zone between roughly 61.8% and 65% is often called the "golden pocket".
78.6% - a deep pullback. Beyond this, many traders would question whether the original move is still intact.
How to draw it, step by step
- Identify a completed swing. You need a clear starting point and a clear end. If you cannot point to an obvious swing low and swing high, there is nothing meaningful to measure.
- Pick the direction. For an upswing that is now pulling back, drag from the swing low to the swing high. For a downswing that is now bouncing, drag from the swing high to the swing low. The tool needs to know which end is the origin of the move.
- Choose wicks or bodies, then stick with it. Most traders anchor to the wick extremes. Bodies are defensible too. Mixing the two between charts makes your levels inconsistent.
- Check that 0% and 100% landed where you meant. Platforms label the ends differently. Make sure the level marked 100% (or 0%, depending on the platform) is the origin of the move, so the 61.8% line sits where a deep pullback would be.
The hard part: which swing
Every chart has many swings, and each one produces a different set of levels. This is where Fibonacci retracement goes wrong for most people, and it goes wrong quietly.
If you are allowed to choose any swing, you can almost always find one whose retracement levels line up with wherever price happened to turn. That feels like confirmation. It is actually hindsight: you chose the measurement after seeing the outcome.
Some ways to keep it honest:
- Use the most obvious swing on the timeframe you are analysing. If someone else looked at the same chart, would they pick the same two points?
- Decide before price reaches the level. Draw the retracement while the pullback is in progress, not afterwards.
- Prefer higher-timeframe swings. A swing visible on the daily chart is one many participants are measuring. One that only exists on the 3-minute chart is not.
Why the levels are not magic
There is a lot of writing that treats Fibonacci levels as natural laws markets obey. That framing is not supported by anything a chart can show you. Price pulls back by all sorts of amounts, and with five or six levels drawn across a swing, price will always be near one of them.
Where the levels can matter is as a shared reference point. Enough traders draw retracements on prominent swings that orders cluster around popular levels, which can make them behave a bit like any other widely watched price. That is a crowd behaviour effect, and like all such effects it is inconsistent.
Confluence is the whole point
A Fibonacci level on its own is one line among several. It becomes interesting when it coincides with something you would have marked anyway.
| Fib level lines up with... | Why it adds weight |
|---|---|
| A prior support or resistance zone | The level is already proven by past price behaviour, independent of the Fib tool. |
| A rising or falling trendline | Two different methods point to the same area. |
| A round number | Weak on its own, but another reason for orders to cluster. |
| A higher-timeframe level | More participants can see it. |
If the 61.8% line sits in empty space with nothing else around it, there is not much reason to expect anything special to happen there. If it sits on top of an old support zone that has held three times, the Fib tool has just confirmed something you could already see.
A worked example, in words
Suppose a stock rallies cleanly from a swing low at 80 to a swing high at 100 over several weeks, then starts to pull back. You drag the tool from 80 to 100. The levels land at roughly 95.3 (23.6%), 92.4 (38.2%), 90 (50%), 87.6 (61.8%) and 84.3 (78.6%).
Now look left. Before the rally, price spent a month consolidating with a ceiling around 88. That old ceiling, if it holds as support on the way back down, sits right next to the 61.8% level. That is confluence: two independent reasons to pay attention to the 87 to 89 area. A pullback that stalls there and shows rejection is more interesting than one that stalls at 92.4 in empty space.
Conversely, if price falls cleanly through 84 and the old consolidation without pausing, the pullback has become deep enough that the original rally's structure is in question. The levels did not "fail"; they were never promises. They were a way of organising where to look.
Common Fibonacci mistakes
- Measuring a swing that is not finished. If price is still making new highs, the 100% point keeps moving and so do all the levels.
- Picking the swing after the fact. Choosing the anchor points that make a past turn line up perfectly is hindsight, not analysis.
- Drawing too many retracements at once. Several overlapping Fib grids from different swings will put a level at nearly every price, which tells you nothing.
- Treating the levels as exact. Price respecting "the 61.8%" usually means price reacted somewhere near it. Read the levels as zones.
- Ignoring the timeframe. A retracement of a 20-minute swing and one of a six-month swing are not equally significant, even if they share a ratio.
Extensions, briefly
Fibonacci extensions project levels beyond the end of the original swing - 127.2%, 161.8% and so on - and are used by some traders as possible targets if a trend continues. They carry the same caveats as retracements, with an extra one: they point into price territory where there is no past behaviour at all, so there is nothing to confirm or refute them until price gets there.
What an AI read can and cannot do with Fibonacci
If you have already drawn a retracement and it is visible in your screenshot, an AI read can see where price sits relative to your levels. If you have not, a tool reading the image has to make the same judgement call you would: which swing to measure. That choice is subjective, and two reasonable analysts can pick different swings on the same chart.
It is also worth being careful with any precise level a tool reports. A model reading pixels has to infer prices from the axis, and the exact value of a 61.8% line depends on reading both swing extremes accurately. Treat specific numbers as approximate, and check them against the price axis yourself. The failure modes article covers this kind of invented precision in more detail.
Frequently asked questions
Do you draw Fibonacci from high to low or low to high?
From the origin of the move to its end. For an upswing that is pulling back, drag from the swing low up to the swing high. For a downswing that is bouncing, drag from the swing high down to the swing low. After drawing, check that the deep levels such as 61.8% sit where a deep pullback would be.
What is the most important Fibonacci level?
61.8% is the most widely watched, and the area around 61.8% to 65% is often called the golden pocket. 50% is also heavily watched even though it is not a true Fibonacci ratio. No single level is reliable on its own; a level matters far more when it coincides with prior support or resistance.
Do Fibonacci retracements actually work?
They work as a consistent way to measure pullbacks and as a shared reference many traders look at. They do not work as natural laws that price obeys. With several levels drawn across a swing, price will always be near one of them, so a level only becomes meaningful when other evidence points to the same area.
Should I use wicks or candle bodies for Fibonacci?
Wicks are the most common choice because they capture the true extremes of the swing. Bodies are also defensible. What matters is consistency - switching between them chart to chart, or to make the levels fit, produces levels that only exist because you wanted them to.
Related Articles
- How to Draw Support and Resistance That Actually Holds Up
- How to Draw Trend Lines That Mean Something
- Multi-Timeframe Analysis: How to Pick Your Timeframes and Read Them Together
Scope
This article is educational and is not financial, investment, or trading advice. Nothing here is a recommendation to buy, sell, or hold any security, cryptocurrency, currency, commodity, or derivative. Technical analysis describes what a chart has already done; it does not predict what it will do, and every pattern described here fails a meaningful share of the time. Trading involves risk of loss. Do your own research and consult a licensed financial professional before making any trading decision. ChartCheck is made by the author of this site.