Why Use Multiple Exits?
A recent message from one of our members questioned our use of multiple exits
and the fact that the exits in a particular system were very complex and would
sometimes move closer to the prices and then suddenly move farther away. The
member questioned whether the exits were working properly and wondered about the
logic of having so many different exit strategies operating within one system. I
sent the member a brief reply and promised to write a Bulletin that explained
our philosophy and procedures about the use of multiple exits in more detail.
When we develop trading systems the entry is usually just a few lines of code but the exit strategies and coding are often very complex. We may have a system with only one very simple entry method and that system may have a dozen or more exit strategies. The reason for devoting so much effort and attention to achieving accurate exits is that over our many years of trading we have come to appreciate both the importance and the difficulty of accurate exits.
Entries are easy. Before we enter any trade we know exactly what has occurred up to that point and if those conditions and events are satisfactory according to the rules of our system we can generate a valid entry signal. Entries are easy because we are able to set all the conditions and the market must conform to our rules or nothing happens. However, once we have entered a trade anything can happen. Now that we are in the market the possible scenarios for what might happen to our open position are endless. It would be extremely na´ve to expect to hope to efficiently deal with all possible trading events with only one or two simple exit strategies. However, that seems to be the common practice and, in fact, many popular trading systems simply reverse the entry rules to generate their exits.
We believe that good exits require a great deal of planning and foresight and that simple exits will not be nearly as efficient as a series of well planned exits that allow for a multitude of possibilities. Our exit strategies need to accomplish a series of critical tasks. We want to protect our capital against any catastrophic losses so we need a dependable money management exit that limits the size of our loss without getting whipsawed. Then if the trade is working in our favor we would like to move the exit closer so that the risk to our capital is reduced or eliminated. As soon as possible we need to have a "breakeven" exit in place that prevents our profitable trade from turning into a loss.
In most of our systems, our goal is to maximize the size of our profit on each trade so we do not simply take a small profit once we see it. This goal means that we need to implement an exit strategy that protects a portion of our small profit while allowing the trade to have the opportunity to become a much bigger profit. If the trade went in our favor every day the exits could be greatly simplified but unfortunately that is not the way markets typically trade. We have to allow room for some minor fluctuations on a day to day basis. In order to facilitate our objective of maximizing the profit of each trade, in some cases we may decide to move our exit point farther away to avoid getting stopped out prematurely. For example, lets look at our Yo Yo exit that is based on the theory that we never want to stay in a position after a severe one-day move against us. (See Bulletin number 14 for an explanation of the Yo Yo exit.)
This highly efficient exit is based on measuring the amount of price movement from the previous day's close. For example we may want to exit immediately if the adverse price movement reaches one and a half Average True Ranges from the previous close. This volatility-based exit will move away indefinitely as the result of a series of adverse closing prices caused by days where the price moved against us but our volatility trigger was never quite reached. Obviously an exit that can move away from prices indefinitely is no use at all in limiting the size of our losses so the Yo Yo exit must always be used in conjunction with other exit strategies that do not move away. Now that we have implemented the Yo Yo exit to protect our trade from a severe one-day reversal in direction, we have still not addressed the question of taking profits. So far, we have exits in place to protect from large losses, to lock in a break-even point and to get us out on a sudden trend reversal but we still have not addressed the important issue of taking some profits on the trade.
We like to shoot for big profits and the bigger the profits become the closer we like to protect them. This strategy calls for multiple profit-taking exits. If we have a $1,000 profit we might want to protect 50% of it and be willing to give back $500 of our open profit. We can place an exit at $500 above our entry price. This will allow us to hold the position in the hope that the profit will grow. However if we have a $10,000 open profit I'm sure we wouldn't want to give back 50% of that. Also, let's hope that our exit stop is not still sitting back there at $500 above our entry price. For best results our exits need to adjust at various levels of profitability.
Many traders have asked us about the robustness of a system that has a many exit rules. The general perception is that a system with fewer rules is likely to be more robust. However I would disagree with applying that common belief without careful thought. Look at the exits in these two over-simplified systems:
Use a $1500 money management stop. (Limits loss to $1500.)
When profit reaches $5,000, exit with a stop at entry plus $4500.
Use a $1500 money management stop. (Limits loss to $1500.00)
When profit reaches $1,000, exit with a stop at entry price.
When profit reaches $2,000, exit with a stop at entry plus $1,250.
When profit reaches $3,500, exit with a stop at entry plus $2,500.
When profit reaches $5,000, exit with a stop at entry plus $4500.
When profit is greater than $7,500 exit with a stop at the previous day's low.
Some system traders might argue that since system A has fewer rules it should be more robust (most likely to work in the future.) We would suggest that system B is much more likely to work in the future even though it has more rules. System A is not going to make any money at all if the open profit never reaches $5,000. Once the profit exceeds $5,000 the only exit is at the $4,500 level. System A is very limited in what it is prepared for. It either makes $4,500 or it loses $1500.
As you can see, system B is obviously prepared for many more possibilities. It is conceivable (but not likely) that system A may somehow produce better test results on a historical basis because of an accidental (or intentional) curve fit. However, we would much rather trade our real money with system B. Simpler is not always better when it comes to exit planning.