Chandelier Exit Formula: When the Line Moves Down

The chandelier exit hangs a multiple of Average True Range below the highest high of the last n bars. TradingView's help page prints it as "Long Exit = Highest Price − ATR × Multiplier", with "Short Exit = Lowest Price + ATR × Multiplier" for shorts.

The published guides leave out that the long level can fall. It drops when ATR expands faster than the high rises, and it drops again when the bar holding the old high leaves the lookback window.

What is the chandelier exit?

A stop level built from two numbers: the highest high (or lowest low) over a lookback window, and ATR, which measures how far price has been travelling per bar.

TradingView, which ships it as a built-in indicator, calls it "a volatility-based indicator that identifies stop-loss exit points based on local price extremes and Average True Range (ATR)" and credits it as "created by Charles Le Beau and popularized by Alexander Elder".

It is an exit tool rather than an entry tool, and two guides say so plainly. Corporate Finance Institute: "Most professional traders recommend using Chandelier Exit as a stop loss tool rather than as a tool for generating trading signals." TrendSpider calls it "commonly used as a trailing stop-loss tool rather than generating trading signals".

The attribution is not settled. The inventor's name appears as Chuck LeBeau, Chuck Le Beau, Charles Le Beau, Charles (Chuck) Le Beau and Charles LeBeau. WallStreetMojo credits it to a 1992 book, "Computer Analysis of the Futures Markets", "co-authored with David W. Lucas". Corporate Finance Institute and Stonehill Forex both name Alexander Elder's "Come into My Trading Room", 2002, as the book that carried it to traders.

What is the chandelier exit formula?

Long exit = highest high over n bars - (ATR x multiplier). Short exit = lowest low over n bars + (ATR x multiplier).

Three numbers decide where the level sits, and TradingView's page names them as three separate inputs:

• Length: "The number of bars back to search to determine the highest high for the long exit level and the lowest low for the short exit level."
• ATR length: the lookback used for ATR itself.
• ATR multiplier: how many ATRs below the highest high the level is placed.

A fourth input on that page, Timeframe, chooses which bars feed the calculation rather than where the level sits.

Most write-ups collapse the two lookbacks into a single n, and some say it outright. Corporate Finance Institute: "the Chandelier exit will calculate the highest high or the lowest low for the same period used to calculate ATR." TrendSpider: "The Chandelier Exit calculates the highest high or lowest low for the same period used to calculate the ATR."

WallStreetMojo's first worked example uses two different lookbacks, "In the last 20 days, the highest price reached by XYZ Inc. was $120" alongside "The 14-day ATR for XYZ Inc. is calculated to be $5", while its second example uses 10 for both. So "22 and 3" does not tell you which 22 you are looking at.

True range itself, and which average "ATR 14" uses, is a separate subject.

A worked example, with round numbers

• 22-bar high: 110
• ATR: 2.50
• Multiplier: 3
• Long exit = 110 - (2.50 x 3) = 102.50

ProTraderDashboard publishes the same arithmetic, "Chandelier Exit: $110 - ($2.50 x 3) = $102.50", and a multiplier ladder on those numbers: "2x ATR stop: $105 (tight)", 3x at $102.50, "4x ATR stop: $100 (wide)". Both check out. Benzinga runs the same sum on a currency pair: a 22-day high of 1.20 with "the ATR(22) is 0.01" and a multiplier of 3 gives 1.17.

Keep 110, 2.50 and 3 in mind.

Does the chandelier exit ever move down?

Yes, in two ways, and both follow from the formula everyone prints.

The first is volatility expanding faster than the high rises. Write the level as Exit = H - (m x ATR). It rises only when the rise in H beats m times the rise in ATR. Take the example: the 22-bar high goes from 110 to 112 while ATR goes from 2.50 to 3.50. The new level is 112 - (3 x 3.50) = 101.50, a point lower than before.

TradingView lists, as a condition for the long exit to increase, "The highest price increases by a greater amount than the increase in ATR". Here the high rose 2 and ATR rose 1, so that condition is satisfied, and the level still fell, because at m = 3 the comparison that decides it is 2 against 3. The multiplier is missing from the printed condition. The same page's second increase condition, "The ATR decreases by a greater amount than the decrease in the highest price", does hold whenever the multiplier is 1 or more, which covers every multiplier published below.

The second way down is the old high rolling out of the window. The lookback is the number of bars back to search, so the high is a rolling maximum. When the bar holding 110 drops out of the 22-bar window and the highest high still inside it is 106, the level becomes 106 - 7.50 = 98.50. That is four points lower with no change in volatility.

That is what a rolling maximum minus a changing average does. Few guides say so, and some say the opposite:

• ProTraderDashboard writes that "the chandelier never rises on its own", and that it only moves up when the highest high moves up.
• Benzinga tells the reader "you can adjust your stop loss upward by recalculating the chandelier exit using the new 22-day high and ATR(22)".
• Market-Bulls says "the Chandelier Exit hangs below the high, rising with the price", and that "The Chandelier Stop Indicator moves with the price during a trend but stays still during a reversal".

Two pages get close from the other side. TradingView describes the levels "expanding or contracting the trailing distance in response to changes in volatility or local price extremes", and WallStreetMojo says "During periods of higher volatility, the Chandelier Exit widens the distance between the stop-loss and the current price". A wider distance under an unchanged high is a lower level.

Can you leave it in the market as a resting stop order?

Not as a straight copy of the line. Benzinga says to "Place your stop-loss order just below (for long positions) or above (for short positions) the chandelier exit level" and, in its worked example, to move it only upward. Those are the same instruction only while the line is rising, and it does not always rise.

A level that falls is information about volatility, or about a high ageing out of the window. It is not an instruction to widen a stop that is already live, because moving a resting stop further from entry raises the loss you accepted when you opened the trade.

How your platform handles the mechanics matters, because the indicator and the order are two different things. On ChartLabs the order types are Market, Limit and Stop, and there is no trailing-stop order type: you move a stop or target during a replay by dragging its line on the chart, and the change is saved at once. A level cannot be dragged past the current price, and closed trades are locked. So following a chandelier line is a decision you make bar by bar, and the record shows what you decided.

What are the best chandelier exit settings?

There is no single answer in the published record, and the numbers disagree. Each one keeps the page that printed it.

Setting	Printed as	Source
No default at all	Three inputs documented, no value given for any of them	tradingview.com
22 and 3	"N is the default unit period of 22", multiplier "the default 3.0"	corporatefinanceinstitute.com, 26 March 2019
22 and 3	"Standard Settings: 22-period high, 22-period ATR, 3x multiplier"	protraderdashboard.com
22 and 3	"usually set to 22 days by default", the multiplier "set at three"	market-bulls.com, 10 March 2024
9 and 2.5	One MT4 build's shipped defaults: "The default setting is 9, but we changed it to 22", "The default setting is 2.5, but we changed it to 3"	stonehillforex.com
10, 22, 50	"Short-term (10 periods): More responsive, exits trends earlier", 22 balanced, "Long-term (50 periods): Very patient, stays in major trends"	protraderdashboard.com
22, 30, 50, and 14 to 30	"Common periods are 22, 30 or 50", and periods that "range from shorter periods, like 14 days, to longer periods, like 30 days or more"	benzinga.com, 9 October 2024

22 and 3 is what most write-ups call standard, original or default, and two of them give the same reason for the 22: "there are usually 22 trading days in a month" (Corporate Finance Institute) and "ATR(22) was chosen because there are 22 trading days in a month" (Stonehill Forex). Stonehill Forex, writing about an actual MT4 build, reports that it shipped with 9 and 2.5, and says it changed those itself "out of observance to the original design". TradingView, which ships the indicator, documents the three inputs and gives a value for none of them.

So the settings rule is simple: record both lookbacks and the multiplier next to every reading you write down. The same bar gives a different level under a different pair of lookbacks, and "22 and 3" does not say whether the 22 belongs to the high, to ATR, or to both.

Chandelier exit against three neighbors

Four stop constructions, each described from its own page. What fits whom depends on whether you want the level to be able to loosen.

Tool	How the level is built	Can the plotted level loosen?
Chandelier exit	Highest high over n bars minus ATR x multiplier (TradingView)	Yes, by construction: the published formula has no step that stops it
Chande Kroll stop	"Chande Kroll Stop = Highest High – (ATR * Factor)", credited to "Tushar Chande and Stanley Kroll" (WallStreetMojo)	The formula as printed there is the same shape, so the same two cases apply
Supertrend	Bands from (high + low) / 2 offset by multiplier x ATR, then "upperBand = basicUpperBand < prev upperBand or prev close > prev upperBand ? basicUpperBand : prev upperBand" and the mirror for the lower band (TradingView)	No, until price closes through it: that update rule only accepts a tighter band
Percentage trailing stop	"Percentage stops use a fixed percentage, while the Chandelier Exit adapts to volatility" (ProTraderDashboard)	No, if you only ever ratchet the percentage

That Supertrend update rule is the whole difference. Its published construction keeps the band from loosening. The chandelier exit's published construction does not.

Where do I get it?

TradingView ships it as a built-in indicator and documents the three inputs quoted above. Stonehill Forex describes an MT4 build, "The MT4 version came to us in 2008", with four settings where the original had two, and gives it away from its own indicator library. Market-Bulls points readers at a community script on TradingView.

Chandelier Exit is one of the standard indicators on ChartLabs charts, under Standard in the Indicators menu. Our defaults are Length 22 and ATR Multiplier 3, with Use Close for Extremes off, so the extremes are the highest high and the lowest low. One Length sets both the extreme window and the ATR period. It draws the long and the short stop, as LeBeau published them, and neither is ratcheted, so the long stop moves down when an old high leaves the window or the ATR widens. Nothing backtests it for you, so the testing stays manual.

How do you test it in a backtest?

With your own run, because the published record has no tested number in it to borrow. Stonehill Forex did run a test, on EUR/USD, BTC/USD, XAU/USD and SPX500 on the daily and 4-hour charts in the MT4 strategy tester, measuring total trades, win/loss ratio and ROI. Its results sit in an image under the words "Below is the data from our testing", and its text gives no figure, only "Numbers were a bit on the low side for the EUR and SPX500".

If you want your own, record these per trade, or the comparison will not mean anything:

• the high lookback and the ATR lookback, separately
• the multiplier
• the symbol and the bar interval
• whether the level fell during the trade, and whether you honored it or left your stop where it was

That last line is the one most logs leave out, and it is the one that decides whether you tested the indicator or tested a one-way version of it.

ChartLabs is a manual backtesting app: you replay historical bars and place practice trades by hand, which is where a level that moves both ways is worth watching bar by bar. Each trade takes notes and tags, and the notes come through in the CSV export, so a note that the level fell and you left the stop where it was is something you can read back later.

If you are new to the process, start with our beginner's guide to backtesting a strategy, or the forex backtesting guide if you are testing on currency pairs. Backtesting accuracy covers the spread and slippage assumptions that decide whether a stop test is realistic.

Next step

Fix both lookbacks and the multiplier, then step one symbol bar by bar and log every time the level fell. ChartLabs has a 7-day free trial with no credit card, then one plan at $20 a month or $200 a year: start your free trial.