Fibonacci Retracement Strategy: A 3-Year Backtest on 42 Crypto Perps

Updated: 2026-10-22 · Türkçe

Fibonacci retracements are one of the most used tools in technical analysis, yet whether they work is rarely measured. We tested one simple, fixed rule on ~3.4 years of 42 Binance crypto perpetuals with 0.1% fees and compared it with RANDOM entries using the same risk/reward. The answer depends on the timeframe: a clear edge on the 4-hour chart, none on the hourly.

Levels used

Instead of the classic 0.236 / 0.5 / 0.618 set we used bands: a 0.34–0.382 shallow band, a 0.618–0.66 deep band and 0.786 as the last defence. The Fib is drawn on wicks from the older end of the leg to the newer; the leg is the highest and lowest point of the last 110 days.

The rule

Kept deliberately simple and unchanged throughout the test:

LONG only: up-leg (low first, then high) and the daily close above its 50-day average
Leg at least 6 ATR
Entry: limit buy at 0.382; one entry per leg
Stop: 0.3 ATR below 0.618
Target: the leg high; time exit after 15 days on 4H

Results (fees included)

Profit factor = gross profit / gross loss. "Random" uses the same stop/target distances at random times — it shows whether the level itself adds anything.

Chart	Trades	Win rate	Avg R	Profit factor	Random entry	Worst year
4-hour	443	56%	+0.28R	1.84	1.29	1.44
Daily	447	48%	+0.12R	1.27	0.99	0.99
1-hour	1,063	44%	0.00R	1.01	—	—

What we learned

On the 4-hour chart the 0.382 entry clearly beats random entries (1.84 vs 1.29) and stayed profitable every year. Daily is weak; hourly has no edge.

Rejected variants: shorts (4H profit factor 1.17, unstable), deep-band entries (negative on daily) and a tight stop just under 0.382 (20% win rate, profit factor 1.18).

Adding a liquidation condition

Requiring heavier estimated long liquidations below price (score < −0.3) and price not below a statistical lower band cut 4H trades to 264 and raised the win rate to 61% and profit factor to 2.25 (Fib trades failing the condition: 1.33). On the daily chart the combined rule was worse than plain Fib (1.16 vs 1.30).

Limits

Pairs were chosen by today's volume (survivorship bias favours longs — hence the random-entry baseline). Funding and slippage are not included. Past results do not guarantee the future.

Free tools

Open the chart (Indicators → Fib Zone Plan) →Liquidation levels guide →

Frequently asked questions

Does a Fibonacci retracement strategy work?

In our test it depends on the timeframe: on 4H the 0.382-entry rule had a 1.84 profit factor over 443 trades (random entries 1.29); on 1H there was no edge (1.01). Not investment advice.

Which Fibonacci level is best?

Here 0.382 (shallow band) worked better as an entry than the 0.618–0.66 deep band; 0.618 served as the stop reference.

Which timeframe is best for Fibonacci?

Of the three we measured, 4-hour was strongest, daily weak and 1-hour had no edge. Others were not tested.

How do I see these levels on a chart?

On the ByStock chart, Indicators → "📐 Fib Zone Plan" draws the bands automatically with entry, stop, target and that timeframe's test result. Free.