Chandelier Exit: Formula, Settings and a Test of 2, 3 and 4 ATR

The Chandelier Exit is a trailing stop that hangs a multiple of the average true range below the highest high of recent bars, like a chandelier from a ceiling. StockCharts' ChartSchool credits it to Charles Le Beau and notes that it was featured in Alexander Elder's books. With the default settings, the exit for a long position is the 22-bar high minus 3 times the 22-bar ATR; the exit for a short is the 22-bar low plus 3 ATR. Below: the formula, the chart on real Bitcoin bars, and a test of multipliers 2, 3 and 4 as the exit of the same entry rule on three series.

Chandelier Exit formula

The line for longs rises when the price makes new highs and falls when volatility expands, because the ATR term grows. That second point surprises some traders: unlike a strict trailing stop that only moves up, the chandelier line can move down while you are in a trade if the highest high of the window drops out or the ATR jumps. How the ATR itself is computed is on average true range.

How to use the Chandelier Exit

• As an exit for a long. Stay in while the close is above the long line; exit when it closes below. This is the use the indicator was built for.
• As an exit for a short. The mirror: stay short while the close is below the short line.
• Choosing the multiplier. ChartSchool's default is 3. It says more volatile stocks need a bigger buffer, so the multiplier should be increased, and shows a 2012 Hewlett-Packard downtrend with a multiplier of 5. A smaller multiple exits earlier and more often.

The Chandelier Exit does not tell you when to enter. It is often paired with a breakout rule, as in the test below, where the entry is a close above the upper line of a 22-bar Donchian channel. A related ATR stop built on two windows is the Chande Kroll Stop, and the general idea of a stop that follows the price is covered on trailing stop.

Testing the multiplier: 2, 3 or 4 ATR

Rule, fixed in advance: when flat, buy at the close when the close is above the highest high of the previous 22 bars; exit at the close when the close falls below the chandelier line for longs (22 bars, multiplier 2, 3 or 4). Long only, no costs, one position at a time. Series: Bitcoin daily from March 8, 2025 to October 28, 2026, the S&P 500 ETF daily from January 4, 2021 to October 28, 2026 (dividend-adjusted), and EUR/USD daily from March 15, 2021 to June 30, 2023. Give-back is the average distance between the best close during a trade and the exit, in percent of the entry price.

Open trades on October 28, 2026 (June 30, 2023 for EUR/USD), not counted above: with a multiplier of 4, Bitcoin long since July 21, 2026 at 66,556, up 22.2% at 81,344, and the ETF long since July 10, 2026 at 753.08, up 2.6%; with 3 and 4 on EUR/USD, long since June 15, 2023 at 1.0945, down 0.3%.

One pattern holds on all three series: a wider multiplier gives fewer trades, held longer, and gives back more from the best close before exiting. Whether that pays depends on the market. On the ETF, which rose 126% over the period, a wider stop stayed in the uptrend longer, and the trades compounded to +9.3%, +28.2% and +49.5% for 2, 3 and 4; buying and holding did far better. On Bitcoin, which ended the period 5.7% lower, the multipliers of 2 and 3 lost (−8.3% and −15.3% compounded); the multiplier of 4 made +11.0% on four closed trades, too few to separate skill from luck. EUR/USD lost with every multiplier. With 4 to 38 trades per line, none of these differences is statistically reliable: the table shows the trade-off, not a best setting. This test is separate from the chart pattern test bench, whose method is described in the methodology.

More volatility tools are in volatility indicators.