Fibonacci Retracement Guide

At a glance

Common levels: 23.6, 38.2, 50, 61.8, 78.6 percent
Derivation: Ratios from the Fibonacci sequence, plus 50 percent which is not one
Main weakness: The swing points are chosen subjectively
Honest use: As a way to pre-select pullback zones, not as a prediction

Key takeaways

The levels come from ratios in the Fibonacci sequence, though the widely used 50 percent level is not a Fibonacci ratio at all.
Because the trader chooses which swing high and low to connect, different traders produce different levels on the same chart, which makes objective testing difficult.
Studies that define the levels mechanically generally find performance similar to other pullback measures such as a percentage retracement or a moving average.
The 38.2 to 61.8 percent zone overlaps with where pullbacks commonly end for reasons unrelated to Fibonacci, which explains much of the apparent success.
The defensible use is as a pre-defined pullback zone within a trend, combined with a trigger and a stop, rather than as a standalone signal.

How the levels are constructed

Fibonacci sequence: 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89 ...

Ratios between terms converge:
   Each term / next term        -> 0.618
   Each term / term after next  -> 0.382
   Square root of 0.618         -> 0.786
   1 - 0.618                    -> 0.382

Retracement levels on a move from 100 to 150 (a 50 point advance):
   23.6%  ->  150 - 11.80 = 138.20
   38.2%  ->  150 - 19.10 = 130.90
   50.0%  ->  150 - 25.00 = 125.00   (not a Fibonacci ratio)
   61.8%  ->  150 - 30.90 = 119.10
   78.6%  ->  150 - 39.30 = 110.70

Extensions project beyond the move:
   127.2%, 161.8%, 261.8% of the original swing

The ratios and how they are applied to a price swing.

Note that 50 percent, one of the most commonly cited levels, does not appear in the Fibonacci sequence at all. It is included by convention because a half retracement is a natural reference point, which is itself a hint about why these levels seem to work.

Why they appear to work

The zone is broad. Between 38.2 and 61.8 percent covers a quarter of the entire move. Pullbacks frequently end somewhere in that range for ordinary reasons.
Self-fulfilment. The levels are widely watched, so orders cluster near them, which produces genuine reactions independent of any mathematical property.
Overlap with other references. Fibonacci levels frequently coincide with moving averages, prior swing points, and round numbers, and it is unclear which is producing the reaction.
Selection of swings. With freedom to choose which high and low to connect, a level can be found near almost any turning point after the fact.
Confirmation bias. The instances where price reversed at a level are memorable; the many instances where it passed straight through are not.

A defensible way to use them

If you use Fibonacci levels, use them to pre-define a zone of interest within an established trend, then require a separate trigger and a defined stop. This converts a subjective drawing exercise into a testable rule.

Swing definition: Mechanical: the highest high and lowest low over a fixed lookback, for example 50 bars. This removes the discretion that makes testing impossible.
Trend filter: Price above a rising 100-period moving average for long setups.
Zone of interest: The area between the 38.2 and 61.8 percent retracements of the most recent defined swing.
Trigger: A close back above the prior bar high while price is within the zone. The zone alone is not an entry.
Stop: Below the 78.6 percent level or 2 x ATR below entry, whichever is closer.
Target: The prior swing high, then a trailing exit for the remainder.
Position size: Risk 0.5 percent of equity based on the stop distance.

Tested this way, the strategy is a pullback-continuation strategy. It may well be profitable, but the profitability comes from buying pullbacks in trends, which is a documented effect. Whether the specific ratios add anything beyond a generic percentage retracement is the question, and the honest answer from mechanical testing is usually that they do not.

Extensions and projections

Extensions project levels beyond the original move, commonly at 127.2, 161.8, and 261.8 percent, and are used as profit targets. The same caveats apply, with an additional one: a fixed target caps the tail of a trend strategy, which is where much of its expectancy lives.

Extensions are reasonable as partial-profit levels within a scaling-out plan, where they serve as any structured target would.
Using them as full exits in a trend strategy reduces expectancy for the reasons discussed in take profit strategies.
The measured-move convention, projecting the prior swing height from the breakout, produces similar levels with a clearer rationale.
Harmonic patterns such as Gartley and Butterfly build on these ratios and inherit all the same definitional problems, multiplied by having more points to fit.

Frequently asked questions

Do Fibonacci retracements actually work?

When tested with mechanically defined swing points, they perform similarly to other pullback references such as a fixed percentage retracement or a moving average. The reactions that do occur are plausibly explained by the levels being widely watched and by their overlap with other references. They are a way of organising expectations rather than a source of edge.

Why is 50 percent included if it is not a Fibonacci number?

By convention. A half retracement is an intuitive reference and was used by technical analysts before Fibonacci ratios became popular. Its inclusion is a useful reminder that the levels are partly a convention rather than a mathematical property of markets.

Which swing points should I use?

For testing, a mechanical definition such as the highest high and lowest low over a fixed lookback. For discretionary use, the most recent clear impulse move. The important discipline is choosing before you look for a level, because choosing afterwards guarantees you will find one that fits.

Is the 61.8 percent level special?

It is the most watched of the ratios, which gives it a modest self-fulfilling quality in liquid markets. There is no evidence that it has a property beyond that. A deep pullback to around two thirds of a move is a common feature of trends regardless of what you call the level.

Should I use Fibonacci in my strategy?

Only if you can define it mechanically and test it against the simpler alternative of a fixed percentage retracement. If the Fibonacci version does not outperform the simple version, use the simple version, because it has fewer discretionary choices and therefore fewer opportunities to fool yourself.

Test this idea before you trade it

Describe the rules in plain language and AlgoTrader AI turns them into a structured strategy blueprint with a configurable historical backtest, cost assumptions, and exportable code.

Build a backtest

Keep reading

Educational use only. This guide explains how a strategy works. It is not investment advice, not a recommendation, and no result described here is a forecast. Test any approach on historical and out-of-sample data, size positions conservatively, and never risk money you cannot afford to lose.