Fibonacci Retracement and Extension Explained: Drawing, Confluence and an Honest View
Lesson 8 of 10 · Intermediate
Where the Fibonacci ratios come from
The Fibonacci sequence (0, 1, 1, 2, 3, 5, 8, 13, 21 and so on) is a series in which each number is the sum of the previous two. As the numbers grow, the ratio of each number to the next approaches about 0.618, and the ratio of each number to the one two places later approaches about 0.382. The common levels used in technical analysis are:
- Retracement levels: 23.6%, 38.2%, 50%, 61.8% and 78.6%.
- Extension and projection levels: 100%, 127.2%, 161.8% and 261.8%.
The 50% level is not a Fibonacci ratio at all; it comes from the older idea of “retracing half of the move.” The 78.6% level is the square root of 0.618, and 127.2% is the square root of 1.618. The link between these ratios and patterns in nature is appealing, but it is no reason for financial markets to obey them.
How to draw Fibonacci levels
- Choose one clear swing. From a meaningful swing low to a meaningful swing high in an uptrend, or the reverse in a downtrend. If you are not sure where the swing starts and ends, the levels will not be reliable.
- Direction of the drawing. In an uptrend you drag the tool from low to high and the retracement levels appear below the high; in a downtrend, from high to low.
- Wick or body. Drawing from the extreme wick points is more common. Whichever you choose, stay consistent across charts.
- Timeframe. Pick a swing that matters on a higher timeframe and study price behavior at the levels on a lower one.
Hypothetical example: gold has risen from 3800 to 3900, a move of 100 dollars. Retracement levels: 38.2% is about 3861.8, 50% is 3850 and 61.8% is about 3838.2. If price makes a low at 3845 and rallies again, the 100% projection of the first leg from the new low (3845 + 100 = 3945) and the 161.8% extension of the first leg (3800 + 161.8 = 3961.8) are the next areas to review.
Confluence: where Fibonacci starts to mean something
A Fibonacci level on its own is not strong evidence. Its value increases when it overlaps with independent evidence:
- a former support or resistance level, such as a previous high that now acts as support;
- a moving average that has repeatedly acted as dynamic support in the current trend;
- a trendline or the midline of a channel;
- Fibonacci levels from two different swings landing in the same area.
In the example above, if the market's previous high was at 3848 and the 50 EMA is near 3852, the 3840 to 3855 zone has multiple confluences. But be careful: the more tools you place on a chart, the more likely it is that “something” overlaps with any given area. Genuine confluence means evidence from truly different sources, not five tools all built from the same price series.
An honest look at the evidence
Fibonacci is one of the most popular tools, but the statistical evidence that its levels outperform random levels is limited and mixed. There are several structural reasons for caution:
- Too many levels. With five or six retracement levels, almost any pullback stops near one of them, and memory keeps only the hits.
- Freedom in choosing points. By shifting the starting point of the swing, you can always find a drawing that “worked” after the fact.
- Self-fulfilling effects. Some reactions may occur because many people place orders there. That effect is real but neither constant nor dependable.
The balanced view: treat Fibonacci as a way to organize possible zones, not to predict a turning point. See levels as zones, wait for price to react in the zone, and if you want to build a rule on them, define the swing in advance and test it on enough data.
Extensions for target management
Extensions are mostly used to estimate possible areas where a move might end, especially when price has reached territory with no price history, for example when gold prints a new all-time high and there is no prior support or resistance. In that situation 127.2% and 161.8% at least provide a framework for scaling out or moving a stop. But reaching an extension does not mean the trend is over; strong trends sometimes pass several extension levels in a row. See an extension as a review point, not a definite end.
The difference between an extension and a projection is worth knowing. An extension is measured on the first swing itself and shows levels beyond 100%; a projection re-measures the length of the first swing from the end of the pullback (the new low). If both land in one area, for example the 100% projection and the 127.2% extension within a few dollars of each other, that area gains importance for review.
Fibonacci in practice: retracement depth as information
One of the less-discussed uses of Fibonacci is gauging trend strength rather than finding an entry. Shallow pullbacks, say to about 23.6% or 38.2%, are typically seen in strong trends where buyers are unwilling to wait for lower prices. Deep pullbacks, say to 61.8% or 78.6%, show that the other side was able to take back a large part of the move and the quality of the trend is lower.
A pullback that passes through 78.6% and threatens the start of the swing usually means that swing is no longer a good reference for drawing and market structure is changing. In that case, rather than hunting for the next level, it is better to read the new structure from scratch.
Common mistakes
- Shifting the swing start and end points until the level fits price.
- Treating a Fibonacci level as an exact price instead of a zone.
- Drawing Fibonacci on a swing whose end is not yet clear.
- Creating fake confluence with many price-derived tools.
- Believing that 61.8% is “golden” and must hold.
What does volume say here?
One of the most useful combinations is comparing Fibonacci levels with the volume profile of the same swing. Suppose the 61.8% retracement sits at 3838 and the volume profile of the up-leg shows the most trading (the POC or an HVN) around 3840. In that area participants have genuinely taken positions and may defend them. This confluence of two different sources, price geometry and the actual location of trading, is more meaningful than the overlap of two price-based tools.
Conversely, if a Fibonacci level falls in the middle of a low volume node (LVN), where the market previously moved through quickly, the chance of price stopping exactly there is lower, and price may continue to the next high volume node. When price reaches the zone, delta and order book behavior, such as absorption of aggressive selling or large resting limit orders, show whether buyers are really present. Volume tools have errors too, but they answer the question “why here?” in a way Fibonacci cannot. For gold, profile and delta come from COMEX GC futures converted to XAUUSD with the basis; CFD volume is tick volume.
📊 See real gold futures volume behind these ideas: free order flow course · today's gold volume map
Frequently asked questions
Which Fibonacci level matters most?
The 38.2%, 50% and 61.8% levels are used most, but no level has a proven edge on its own. A level's importance depends more on its confluence with price structure and high-volume areas.
Should I draw Fibonacci from wicks or bodies?
Drawing from extreme wick points is more common, but consistency matters more than the choice, so results are comparable and testable.
Does Fibonacci have scientific backing?
Statistical evidence for the superiority of Fibonacci levels is limited and mixed. Treat it as a tool for organizing possible zones, and test any rule you build on it yourself.
Education only, not a recommendation to buy or sell; we never give signals. Price examples are illustrative. Test any method on a demo account first, with strict risk management.