The following article was originally published in the December 2007 issue of
The Breakout Bulletin.
Exits for All Occasions
most people think about trading systems, they probably think about how
the system enters the market. In fact, trading systems are usually
described in terms of their entry technique, such as breakout systems,
moving average crossover systems, Fibonacci retracement systems, and
myriad other methods for entering the market. Even the terms
"counter-trend" and "trend-following" refer more to the entry than the
Despite the common focus on entry techniques in trading, you may have
come across the assertion that exits are more important than entries. In
my experience, that's usually true. Why? One possible reason is that most market
action is random. A good trading system finds at least some signal in
all that noise. But with all the noise in the market, a substantial
percentage of entry signals may be wrong.
long-term trend following system, for example, may be right only 40% of
the time. Despite the low percentage of winning trades, it can still be
highly profitable if it keeps the losing trades smaller than the
winners. The way it does that is by cutting the losses short and letting
the winners run. In other words, it's profitable because of how it
handles the exits.
Generally speaking, I think it's fair to say that exits are the
principal method of controlling the intrinsic risk/reward
characteristics of a trading system. Whether the system looks for a
quick profit or holds the trade through the market's ups and downs
depends on the exits. Likewise, whether a losing trade is cut off
quickly or given more room to move is determined by the exits. Exits
are truly the way to implement "cut your losses short and let the
that I use the word "intrinsic" when describing the risk/reward
characteristics of a trading system to differentiate the rules and logic
of a trading system from position sizing. Certainly, position sizing can
be used to improve the overall risk/reward ratio of a trading system,
but position sizing is an external factor, apart from the rules and
logic of the system. The focus of this article is on the rules and logic
of trading systems, rather than on position sizing.
The following list is not
exhaustive but it includes some of the most common types of exits you
may encounter or consider for your own trading systems:
and Reverse. This is basically an "exit-less" exit. Stop and reverse
systems reverse from long to short and back to long again. If you're
long one contract, for example, you would sell two to close out the long
trade and go short. You're never flat the market with this type of exit
because each exit is also an entry in the opposite direction.
Bars from Entry. Exit the trade at the market N bars from the bar of
entry, where N can be any number greater than zero. For example, you
might exit the trade 10 bars from the entry. The duration of the trade
will depend on the bar size; e.g., 5 min bars or 60 min bars.
of Day. Rather than exiting relative to the entry, as with the
previous method, you exit at a specific time of day, such as at 10:30
am. As a special case, this exit also includes exiting at the end of the
Management Stop. This is a commonly used exit type that uses stop
orders to limit the risk of a trade. For a long trade, a sell stop order
is placed below the entry price. For a short trade, a buy stop order is
placed above the entry price. When the stop is hit, the loss is limited
to the size of the stop plus slippage. Common methods on which to base
the size of the stop are a fixed dollar amount, a fraction of the
average true range, or as a percentage of price.
Trailing Stop. This type of exit uses stop orders to lock in a
percentage of the open profit after a specified level of open profit --
the floor -- has been reached. For example, after an open profit of $500
has been reached on a long trade, you might place a stop order below the
market so that 50% (or $250) of the open profit will be retained if the market
reverses. The floor amount is typically computed based on either a fixed
dollar amount or a fraction of the average true range. Note that a
breakeven stop is a special case of the trailing stop where the
percentage of profit to lock in is zero.
Profit Target. The profit target uses limit orders to exit when a
specified price has been reached, representing a profit for the trade.
For a long trade, the limit order is above the market; for a short
trade, it's below the market. Like a trailing stop, a profit target can
help avoid giving back open profits when the market reverses. However,
profit targets also place a limit on the maximum profit that's possible
from a trade.
Logical conditions/trading logic. In addition to the exit types
listed above, just about any logical condition, such as those used for
trade entries, can be used to exit a trade.
Price Patterns. For example, a series of consecutive lower closes
might be used to exit a long trade.
Trend Indicators. For example, moving averages, momentum, and MACD
could all be used to signal a trade exit.
Trend Strength. For example, when the ADX indicator, which measures
trend strength, declines below a certain level, a trend-following
trade might be exited.
Oscillators. Stochastic, %R, RSI, Bollinger bands, and other
oscillators measure over-bought and over-sold conditions. A long
trend trade might be exited when an oscillator crosses below the
over-bought level, indicating the end of an up-trend. On the other
hand, if the trade has been entered on weakness, such as a
pull-back, a short-term trade might be exited when the oscillator
crosses above the over-bought level.
Support/resistance. Support and resistance levels can be used either
as money management stop prices or as target prices.
A Comparison Study
To test and compare
different exit types, a simple system was written in TradeStation's
EasyLanguage. The system, called ExitTests, is available on my
downloads page. ExitTests enters
trades using a 20-bar channel breakout: it buys at the highest high of
the last 20 bars plus 1 point on a stop and sell short at the lowest low
of the last 20 bars minus 1 point on a stop. Channel breakouts tend to
work fairly well as trend following systems.
The following exits were
Stop and reverse.
20 bars from entry.
Money management (MM)
stop: size equal to 1 x average true range (ATR) of the last 20 bars -- ATR(20).
Trailing stop: floor
at 1 x ATR(20); 50% profit lock.
Target: 1 x ATR (20)
from the entry.
Moving average: exit
long when close is below moving average (MA) of last 20 closes --
MA(20); exit short when close is above MA(20).
RSI: exit long when
RSI of the last 14 closes -- RSI(C, 14) -- crosses above 70; exit
short when RSI(C, 14) crosses below 30.
In addition, the following
combinations of exits were tested:
N bars with trail
N bars with target
N bars with MA
N bars with RSI
MM stop with trail
MM stop with target
MM stop with MA
MM stop with RSI
The same parameter values
were used for all tests and all markets. These values, as listed above,
were not optimized.
Each exit was tested over
the following five futures markets: crude oil (CL), E-mini Russell 2000
(ER2), Japanese Yen (JY), T-bonds (US), and wheat (W). The test period
was 20 years of daily data where available (six years for the ER2), and
no commissions or slippage were deducted. The tests were conducted in TradeStation 8.2.
The results are shown in
the following figures. The corresponding Excel spreadsheet is available
online on my downloads page.
Exit test results for 20-bar channel breakout system, single exit types.
Exit test results for 20-bar channel breakout system, combination exit
The figures display the
net profit (Net Prof), worst-case drawdown (DD), and profit factor (PF)
for each market and exit type. The best exit type
for each market is highlighted in green, and the
worst exit type is in red. The best and worst were chosen
subjectively based on a combination of the three performance metrics
(profit, drawdown, profit factor).
It's apparent from the
figures that there is no consistently best exit type. In fact, each
market has a different best exit type. This suggests that it may make
sense to customize the exit rules to the market. However, for some
systems, such as long-term trend-following systems, the number of trades
from one market may be too small for statistical significance. In that
case, it may be necessary to use the same rules for all markets.
Overall, the combination
of N bars from entry with a profit target ("NB-Targ") was the best exit,
as judged by average net profit and average net drawdown. The advantage
of exiting at N bars from entry is that the exit is unaffected by market
volatility. It acts as a way to get out of a trade that isn't working.
It makes sense that this works well with a profit target. This
combination is basically saying "either we hit our profit goal within N
bars or we get out."
The worst exit, somewhat
surprisingly, was the combination of a money management stop and a
profit target ("MM-Targ"). This combination was worse than either the
money management stop or the profit target by itself. The reason for
this is that the profit target was relatively small and the same size as
the stop. This resulted in a large number of small trades, many of which
were stopped out. Without the profit target, the trades were much longer
in duration, and the profits were larger. This illustrates that when a
money management stop and a profit target are combined, they have to be
sized relative to each other. In this case, because the channel breakout
method typically has less than 50% profitable trades, it would be better
to increase the size of the target relative to the stop so that the
average winning trade is larger than the average loss.
Among the single exit
types, the best overall was the target exit, which was also the most
consistent. The worst single exit type was probably the RSI. Considering
that the channel breakout method is a trend-following method, it's
probably not surprising that an oscillator performed poorly as an exit
technique. The trailing stop probably returned the most variable
results. The best results in the Yen and the worst results in crude oil
were both achieved with the trailing stop. This may just mean that the
size of the floor and/or the lock-in percentage for the trailing stop
are market dependent.
One general conclusion
from this study is that the type of exit chosen for a trading system can
make a big difference in the system's performance, even the difference
between winning and losing. How different exits are combined is also
clearly important. Although different results would be obtained with
different entry rules or different markets, hopefully this study will
give you some good ideas for your own trading system development. Just
remember to test every assumption and perform good out-of-sample testing
and/or real-time tracking before committing real money.
That's all for
now. Good luck
with your trading.
SIMULATED PERFORMANCE RESULTS HAVE CERTAIN INHERENT LIMITATIONS. UNLIKE
AN ACTUAL PERFORMANCE RECORD, SIMULATED RESULTS DO NOT REPRESENT ACTUAL
TRADING. ALSO, SINCE THE TRADES HAVE NOT ACTUALLY BEEN EXECUTED, THE
RESULTS MAY HAVE UNDER- OR OVER-COMPENSATED FOR THE IMPACT, IF ANY, OF
CERTAIN MARKET FACTORS, SUCH AS LACK OF LIQUIDITY. SIMULATED TRADING
PROGRAMS IN GENERAL ARE ALSO SUBJECT TO THE FACT THAT THEY ARE DESIGNED
WITH THE BENEFIT OF HINDSIGHT. NO REPRESENTATION IS BEING MADE THAT ANY
ACCOUNT WILL OR IS LIKELY TO ACHIEVE PROFITS OR LOSSES SIMILAR TO THOSE