Option Investor
Trader's Corner

Chandelier Exits

Printer friendly version

Did you ever find a big hole in your education as a trader? I recently did. What I found may prove useful to those whose main trading strategies include directional options, futures or stock trades. This would include buying calls or puts or buying or shorting stocks or futures, for example.

I stick to credit spreads these days, but I use the monthly average true range (ATR) indicator when I'm getting into those spreads, and this new information relates to the ATR. This indicator helps me ensure that I've placed my spreads far enough away from the action that they're not likely to get clipped if the underlying just moseys through its current typical range for a month. I found out that some traders employ the ATR in a specific way for setting stops and not determining entries, however, and that method is what this article will address.

First, a little background might prove helpful to newer traders. According to the Stock Charts' Chart School, Average True Range or ATR was developed by J. Welles Wilder. Its calculation requires several steps, although most charting programs perform these calculations for you automatically. Still, understanding how an indicator is calculated helps in understanding how to best use it. First, those making the calculations find the true range. This is defined as the greatest of three differences: the current high less the current low, the absolute value of the current high less the previous close or the absolute value of the current low less the previous close.

If the SPX has one of its occasional big-range days such as the one on January 25 when its day's range was more than 18 points, then it's likely that the day's high minus the day's low will be the greatest of the three. If the SPX has instead produced an inside-day situation in which one day's candle is inside the previous day's range, then one of the other two calculations is going to be greater and will be used as the true range for that day.

As the name implies, the average true range is then calculated by taking an average of the true range over a certain period of time or number of bars on the chart. The default calculation on my chart provider is 14 periods, but I have substituted 20 on my charts. For example, if I'm looking at a daily chart, the true ranges will be averaged over the previous 20 days.

This indicator speaks to the volatility of prices rather than their likely direction. If prices become more volatile, the ATR rises because all those true ranges that make up the average expand. If prices are rising quickly or even dropping quickly, ATR will rise. Direction of prices doesn't matter: only the width of the calculated true ranges does. If prices become less volatile, ATR declines. Over the last several years, the SPX's monthly ATR has been falling.

Annotated Monthly Chart of the SPX:

In fact, I place the sold strikes of my credit spreads much further than one ATR on the SPX, especially for the bear call spread above the SPX. If a market has been trending as the SPX has been, I would try to get my short strike much further than one monthly ATR away from the current price in the direction of the trend. In other words, at the time this chart was snapped early last week, I wouldn't have wanted to place March bear call credit spreads only 55.80 points away from the then-current SPX price of 1426.61. I would have wanted to be much higher with my sold call.

I'm also lengthening the period over which the true ranges are averaged to 20 periods rather than the default 14 from my charting service because ATR has been trending lower. I want to capture the lengthier period in the averages so that the ATR doesn't get too narrow and I don't inadvertently get too close to the action. This is particularly true because a low ATR can be a sign that a market is topping out.

Think about it. When markets climb, they soar higher and then frequently go through a period when a mushroom or rounding-over-type top is made. While that top is forming, the volatility diminishes as ranges contract. I'm not certain that's happening now, but it remains a possibility. If that should be true, the SPX could plunge faster than it's been climbing. In a falling-ATR environment, I want some of those earlier, higher true ranges averaged in to keep me as far out of danger with my spreads as possible. If and when the SPX plunges lower, I'll shorten the number of periods used in the ATR's calculation so that it will react as quickly as possible to the increasing volatility.

After using the ATR all this time to design my credit spread entries, I happened across an article by Sharon Yamanaka in the November 2006 issue of STOCKS & COMMODITIES describing chandelier exits, a type of exit that employs the ATR in a way that I've never imagined using it. After bullish entries, such as buying a stock or a call, the exit or stop is set by subtracting a multiple of the ATR from the highest high or highest close during the period being considered. In effect, the exit hangs from that highest high or highest close, giving the exit its name. The opposite is done with a bearish entry, with the exit or stop hung from the lowest low or lowest close.

Why would traders want to do that? One benefit of the chandelier exit is that it expands or contracts with the changes in volatility. Yamanaka notes that while she always sets a hard stop upon entering a trade, stops "can do more harm than good" when a too-tight stop takes traders out of a trade too soon or a too-wide one keeps them in a bad trade too long. She believes that this type of exit helps protect against being whipsawed out of what will ultimately turn out to be profitable trades because the exit is wider during volatile periods, when ATR rises. This exit would also tighten when volatility contracts, so that profits are better protected than they would be with an exit that is too wide for the current conditions. The exit or stop can be recalculated at the end of each day so that volatility changes will widen or contract the exit.

Another important aspect of this type of exit, Yamanaka notes, is that it's unlikely to be placed at the same place as the majority of stops might be placed. Stops may be less likely to be overrun in a typical stop-running move meant to take out as many stops as possible.

Sound interesting? It did to me, too. Yamanaka had mentioned that she hadn't been familiar with the chandelier exit, so I wasn't feeling too chagrined at having missed information about this type of exit the whole time I was trading those purely bullish or bearish plays, when I most could have used information about them. However, imagine my surprise when I went looking for other articles on the topic and discovered that a few traders have been advocating chandelier exits for almost a decade. In TRADE YOUR WAY TO FINANCIAL FREEDOM, a book copyrighted in 1999, Dr. Van K. Tharp discussed the methodology, although I'm not certain he employed the term "chandelier exits" when discussing the terminology.

According to Chuck Le Beau and Terence Tan, writing for The Traderclub Forum (December 3, 1999), Tharp hung his exits three ATRs' distance from the highest or lowest close, calculating the ATR with a ten-day exponential moving average. Chuck Le Beau has also written about the chandelier exit in his System Traders Club. In their earlier article, he and Tan advocate an ATR calculated with a 20-period exponential moving average, the same one I use. They suggested keeping the exit wider at the beginning of a trade, perhaps 2.5 to 4 multiples of the ATR, and narrowing it as the trade becomes profitable, perhaps as narrow as one or one-half multiples of the ATR.

Yamanaka sticks with Tharp's ATR using a 10-period average and sets the stop three ATRs away from the highest high on long plays or lowest low on bearish ones. She allows for some adjustment depending on conditions such as trading style, chart formations, time frame and risk tolerance.

All--Yamanaka, Le Beau and Tan--tout the flexibility of these exits and their usefulness in any type of market, whether one is trading the forex markets, corn futures or the QQQQs. An important point to note is that in bullish entries, the chandelier exit will never move lower than the original stop because it's hung from the highest high or highest close, which will change only if there's a new, higher high or close. In a bearish entry, the chandelier exit or stop will never move higher because it's hung on the lowest low or close, and the only way that will change will be if there's an even lower low or close.

Le Beau and Tan note that changing the moving average of the ATR will make the stop at the chandelier exit more or less reactive to current trading conditions. A shorter-term moving average of the ATR will result in stops that react more quickly to changing market conditions. They will sometimes calculate two ATRs, one using their typical twenty-bar calculation and another using a shorter length, perhaps as short as four bars. They'll then use the widest of the two exits calculated using the two methods as their stop. This allows the exit to quickly adjust to expanding volatility without resulting in too many whipsaws when volatility contracts for only a few days before expanding again, a case in which the four-bar calculation might result in a too-tight stop.

Yamanaka also advocates some flexibility. She would study chart patterns and support and resistance levels in conjunction with the calculated chandelier exit or stop. In one example in her article, she calculated a chandelier exit at 3 ATRs from the highest high, putting that calculated exit at $9.77 for her bullish play in her chosen stock. She noted strong historical support at $10.00, so the $9.77 exit made sense to her. One can imagine, though, that if the calculated chandelier exit had been just above strong support rather than just below it, she might have widened her exit so that she wouldn't be taken out just ahead of strong support.

She also cautions that account considerations should be an important part of any consideration of where stops should be. If the calculated chandelier exit is wide, for example, so that the loss would represent too large a percentage of a trader's account, then a smaller number of stock shares or options contracts might be entered than had been originally anticipated. She, like many others, advocates knowing where the stop will be before the trade is entered.

How would this work in practice? An example might be seen on the Russell 2000's daily chart, with this chart captured at the end of the day January 30, when I first began roughing out this article.

Annotated Daily Chart of the RUT:

Note that the stop would also roughly correspond to a retest of the rising trendline that was former support for the RUT as it climbed off the summer lows, so should some extra leeway be given to allow for a complete test of that trendline? What about for the expected volatility that might come after the FOMC decision? Would adding another point or so to that stop be a good idea to allow for that trendline test, especially given the likely post-FOMC-decision volatility? Or would be it better to stick to the 4 times ATR stop or maybe even narrow it further inside their advocated 2.5-4.0 times ATR range?

That would be up to the individual trader and that trader's account and risk tolerance. Some might even have stuck with Yamanaka's tighter three-times-the-ATR stop, although Yamanaka does use chart formations, too, to help her set stops. What would you have decided? Think about it in the context of your trading style and your account size before glancing at the chart below.

Annotated Daily Chart of the RUT:

What decision would you have made? As I type this Friday morning, the RUT is still testing that trendline without a clear result. Because the trendline is rising, Friday's trendline test means that the trendline has moved higher than at Thursday-morning's level. The RUT has been as high as 809.58 this morning, but is at 808 as I type. Perhaps it will print a doji at that resistance today and perhaps it will fall back, but perhaps it will just continue to climb the underside of that trendline.

So was this test a failure? It's possible that those who might have taken a chandelier exit and been stopped would have stood by to watch that play ultimately be profitable for someone else. That doesn't mean that this type of exit is a failure, though. In addition, Yamanaka, Le Beau and Tan all advocate learning the ATR multiple that best works with your preferred trading vehicle, so some experimentation is necessary. There's another possibility, too.

Remember that some advocate basing the chandelier exit not on the highest high or lowest low on the period being considered, but rather on the highest close or lowest close. At the close on 1/30, the date that the first RUT chart was snapped, the lowest low over the last 20 periods had been 774.55, considerably higher than the lowest intraday low. When a chandelier exit or stop is calculated at 4 ATRs' distance from that lowest close, the exit or stop would have been at 812.55, not 806.68. That would have been a wide, wide stop, but it would have allowed for a thorough testing of the rising trendline.

So which stop would have been appropriate? Would one have kept traders from being whipsawed out of a trade that would ultimately be profitable? Would one have protected traders from a bigger loss if the RUT continues zooming higher? That's ultimately still to be proven, and the reason that if you decide that you're interested in these exits, you must experiment.

LeBeau and Tan note that Dr. Tharp conducted a study that employed chandelier exits after random entries were made. According to LeBeau and Tan, Tharp's study concluded that the results were likely to be profitable, even with the random entries. It's of course a study that I have not replicated and can not defend.

I'm not trading the types of trades in which these chandelier stops would be used, either, so I can't tell you about my personal experience with this just-discovered type of trading. I haven't back-tested it. However, I do think it's interesting enough that it's worthy of investigation if you're entering directional plays.

I would suggest that those interested in this type of stop or exit first back test or at least paper test the exit before relying on it for actually setting stops. Most charting services now calculate ATR and most allow you to input the period over which the true ranges will be averaged. If yours doesn't, Yamanaka lists Prophet.net's "Analyze" tab as one source for this indicator's value for your preferred trading vehicle. Terence Tan has written code that allows TradeStation users to code the chandelier exit, with that code available at the www.traderclub.com website.

While I can't guarantee that the chandelier exit measures up to all the claims that have been made about its usefulness, it perhaps warrants a look.

Trader's Corner Archives