Fibonacci Retracement and Extension Levels

What Are Fibonacci Retracement and Extension Levels?

A retracement is a partial reversal of an earlier move. If an illustrative stock rises from $100 to $120, then falls to $110, it has given back $10 of the $20 rise: a 50% retracement.

Fibonacci chart tools mark proportions such as 38.2% and 61.8% of that earlier move. On the same $20 rise, a 38.2% retracement is $7.64 below the high, at $112.36; a 61.8% retracement is $12.36 below it, at $107.64.

The commonly included 50% mark is the halfway point and is not itself a Fibonacci ratio. Extensions use multiples of the move to mark prices beyond its original endpoints.

Choose the Move You Are Measuring

Those endpoints are the anchors. Measuring from a different low gives a different set of levels. Record the choice before examining the later reaction, as you would when drawing a trend line.

• Choose a completed directional move and identify its beginning and end.
• For a rising move, measure from its low to its high; reverse that direction for a decline.
• Keep the timeframe and price scale consistent.
• Retain the original endpoints when examining later prices.

Fibonacci Retracement Levels During a Pullback

In our $100-to-$120 example, a retreat to $115 gives back one-quarter of the rise; a retreat to $105 gives back three-quarters. The percentages describe the depth of the decline. Whether price subsequently recovers is a separate observation.

Traders sometimes compare a retracement mark with an earlier support area. If both are near $110, that gives them one area to watch. A trade at that price still needs an entry and exit rule.

Fibonacci Extension Levels for Targets and Resistance

Using a two-point projection from the $100 low, 127.2% of the $20 move gives $125.44; 161.8% gives $132.36. Both lie above the earlier $120 high. These are illustrative calculations; tools using an additional pullback anchor can place extensions differently.

A trader can use an extension as a planned target. To test that choice, specify whether the trade closes when price reaches it, waits for a close, or follows a trailing exit after arrival. The resulting profit depends on that rule.

When Two Levels Overlap

Confluence means separately drawn levels fall near the same price. A Fibonacci level and a trend line may overlap because they use related turning points. Counting them as independent confirmation can overstate how much new information the second tool adds.

The trendline strategy guide explains how to combine a location with entry, safety-exit and target rules. The research discussion on this page addresses what the saved comparison can tell us about Fibonacci levels themselves.

Checks When Using Fibonacci Levels

• Keep the same anchors when reviewing the later move.
• Avoid ranking a ratio as universally strongest without a test of that claim.
• Define how near two levels must be before calling them an overlap.
• For the $100-to-$120 example, distinguish the calculated $110 midpoint from evidence about what happened after price reached it.
• Specify the entry and safety exit before interpreting a projected target as potential reward.

Do Fibonacci Levels Predict Bounces? Our Study Found a Limit

Read the full research report for the recorded comparison, methods and limitations.

In our saved monthly retracement study, the single Fibonacci-level group had a rebound rate close to the changed-ratio comparison. Overlapping levels did not have a higher recorded rebound rate. The historical anchor labels and incomplete data archive limit what we can conclude.

In plain English: the Fib label alone did not add dependable bounce odds in that test. That does not make the tool useless. It can describe location and trade geometry; it just does not create odds by itself. The statistics below belong to the full setup—context, qualification, entry, stop, and exit—tested together. Read how TradingPal backtests trendlines for the method behind that result.

The sample, instrument scope and dates belong to this exact setup book. Extreme drawdowns remain visible rather than being hidden.

Backtest window: 2021-12-06 to 2026-07-13. Data snapshot: 7/23/2026. Historical results can differ from live fills and do not predict future performance.