Reading the chart 5 / 6
Reading the chart · Level 1Lesson 5 of 5

Fibonacci retracements: how to draw them, and what BTC data shows

Fibonacci levels are on almost every trading chart. Drawing them takes two clicks; knowing what they are worth takes counting. We counted every BTC pullback on three timeframes.

You will learn
  • How retracement levels are drawn on a swing, and where the ratios come from
  • Why a pullback "right at 61.8%" proves little on its own
  • What every real BTC pullback on three timeframes shows about the levels

Before you start: The lessons on trend and market structure, and on support and resistance.

Drawing the levels

A Fibonacci retracement is drawn on one leg of price: from a swing low to a swing high for a rise (or the reverse for a fall). The levels mark how much of that leg a pullback has taken back, measured from the high: 23.6%, 38.2%, 50%, 61.8% and 78.6%.

level = high − ratio × (high − low)

The ratios come from the golden ratio, 1.618…, the number that the ratio of neighbouring Fibonacci numbers (1, 1, 2, 3, 5, 8, 13…) approaches: 0.618 is 1 ÷ 1.618, 0.382 is 0.618², 0.236 is 0.618³ and 0.786 is √0.618. The 50% level is not a Fibonacci ratio at all; it is included by convention.

Fibonacci levels — two real BTC legs on 4-hour candles. From 80,126 on 20 September 2026 to 87,396 on 21 September (+9.1%), the pullback ended at 82,875 on 24 September: 62.2% of the leg, $28 past the 61.8% line. From 80,394 on 8 October to 84,352 on 12 October (+4.9%), the pullback ended at 81,034: 83.8% of the leg, $207 past the 78.6% line, and on 15 October price fell below the start of the leg, to 79,800.

One pullback proves nothing

Five levels between 23.6% and 78.6% sit 12 to 17 points apart. Even if the levels meant nothing, many pullbacks would end near one of them: the bands within 2 points of a level cover a fifth of the whole 0–100% range. To know whether the levels attract price, count every pullback — and compare with levels that mean nothing.

Where pullbacks ended — every pullback of BTC/USDT on three timeframes. 4-hour candles, 28 October 2025 – 27 October 2026: 29 of 116 pullbacks ended within 0.02 of a Fibonacci level; the same five levels shifted by −0.04, −0.02, +0.02 and +0.04 caught 25, 23, 33 and 28. Daily candles, 1 January 2024 – 27 October 2026: 11 of 48, against 13, 11, 13 and 13. Hourly candles, 1 June – 27 October 2026: 41 of 178, against 32, 41, 42 and 50.

What the count shows

On all three timeframes, the Fibonacci levels caught about as many pullbacks as the same levels moved a little up or down — sometimes more, sometimes fewer. One detail stands out: on the 4-hour chart 13 pullbacks ended near 61.8%, more than at any other single level, while on the daily and hourly charts 61.8% caught 2 and 9. With a few dozen pullbacks per level, a cluster like that can appear by chance; it did not repeat on the other two timeframes.

This is one market, one swing rule and one to three years of data, so it is not proof that the levels never matter. It is a reason to treat them like any other level: a place to watch, best when it agrees with something else — a previous swing, a support zone, the value area — and never a price that must hold. Many traders watch the same levels, which can make price react there sometimes; it is just as easy to remember those times and forget the rest.

Check yourself · 1 of 3A leg rises from 80,000 to 90,000. Where is the 61.8% retracement level?
Show the answer

90,000 − 0.618 × 10,000 = 83,820: 61.8% of the rise taken back.

Check yourself · 2 of 3Which of the usual levels is not a Fibonacci ratio?
Show the answer

0.382, 0.618, 0.236 and 0.786 all come from the golden ratio; 50% is included by convention.

Check yourself · 3 of 3How do you check whether Fibonacci levels attract pullbacks?
Show the answer

One good example is easy to find. Only a full count, compared with made-up levels, says whether the real ones do better.

Takeaways
  1. Retracement levels are shares of one leg, measured back from its end: 23.6, 38.2, 50, 61.8, 78.6%.
  2. The ratios come from the golden ratio; 50% does not.
  3. With five levels, a pullback often ends near one by chance.
  4. In BTC data on three timeframes, shifted levels caught about as many pullbacks as the Fibonacci ones.
  5. Treat them as one more level to watch, not as a price that must hold.

Open BTC in the terminal — auto Fibonacci is under the align-left button above the chart

Further reading

Education only, not financial advice · Updated 28 Oct 2026 · Data: Binance · Report a mistake