Exit Logic
The entry gets you in. The exit determines what happens next. Two rules. Two different problems. One keeps the mistake survivable. The other gives the extraordinary trade room to become extraordinary.
Flying a kite requires a peculiar kind of discipline.
You have to maintain enough tension to keep control, but not so much that you prevent the kite from doing what the wind is capable of making it do.
Pull too hard and you kill the flight.
Give it everything and you may lose the kite.
Somewhere between those two lies the art of staying attached while allowing something else to determine how far the journey goes.
There is something of trend following in that.
Exit logic actually contains two quite different mechanisms.
The initial stop answers one question:
How much am I prepared to lose if this entry goes nowhere?
The trailing exit answers another:
How much room am I prepared to give this position if it becomes something extraordinary?
Those questions sound related because both eventually result in closing a position.
But they solve completely different problems.
One deals with failure.
The other deals with success.
And confusing the two can destroy the very payoff structure we are trying to create.
Episode 1 established something fundamental. When an entry fires, the system does not know whether it has detected the beginning of a major trend or another piece of noise.
It enters anyway.
That uncertainty now has to be managed.
The initial stop makes being wrong survivable.
The trailing exit makes being spectacularly right valuable.
Between them, something remarkable happens.
A system that knows almost nothing about the future can produce a profoundly asymmetric return distribution.
That asymmetry is not an accident.
We build it.
The Arithmetic Behind the Asymmetry
The Outlier Hunter lives with an unusual distribution of outcomes.
Lots of small losses.
Some modest winners.
A few very large winners.
And occasionally something that dwarfs almost everything around it.
If you look only at the frequency of those outcomes, the strategy can appear almost perverse. We willingly participate in a process in which being wrong is completely normal.
Sometimes repeatedly.
That is because frequency is not what pays us.
Magnitude does.
Imagine losing one unit of capital five times and then making ten units on the sixth trade.
You were wrong five times out of six.
And you made money.
That simple arithmetic sits at the heart of convex trend following.
It also explains why exit logic matters so much.
If the loss side is allowed to expand, the repeated failures become dangerous.
If the winning side is prematurely constrained, the rare successes can no longer pay for the failures.
So we create two different rules.
The initial stop says:
Enough. This one didn’t work.
The trailing exit says:
Not yet. Let it continue.
One closes quickly when the evidence moves sufficiently against us.
The other deliberately refuses to close merely because we have made money.
That distinction creates the shape of the return distribution.
“Entry logic determines when you get into the market. Exit logic determines whether you are running an outlier programme or an ordinary one.”
The Initial Stop: Making Failure Ordinary
The initial stop has a beautifully narrow purpose.
It defines where we leave if the trade moves sufficiently against us.
Nothing more.
Before we enter, we know the rule. We know the stop distance. We know the intended capital exposure associated with that distance.
There should be no negotiation once the trade is underway.
In an ATR-based system, that distance is related to the recent movement of the market. A market moving violently from day to day needs more price room than one barely moving at all.
This is not because ATR measures risk.
It doesn’t.
ATR measures movement.
We use that movement to scale the stop so that we are not applying the same arbitrary price distance to markets behaving very differently.
That distinction matters.
A 50-point stop might be enormous in one market and meaningless in another. Expressing the distance relative to the market’s recent range gives the system a common language with which to treat very different instruments.
Then position sizing enters the calculation.
A wider stop generally means a smaller position.
A narrower stop generally means a larger one.
The objective is not to pretend that every trade carries identical real-world risk. Markets can gap. Liquidity can disappear. Execution can occur beyond the intended stop.
The objective is to establish a consistent planned loss exposure before the trade begins.
That planned loss matters enormously because trend following generates losing trades as part of its normal operation.
Failure cannot be exceptional.
It has to be affordable.
Again.
And again.
And again.
That is what the initial stop makes possible.
The temptation, of course, arrives when price approaches it.
Perhaps the market only needs a little more room.
Perhaps the signal is still fundamentally sound.
Perhaps tomorrow will be different.
Move the stop further away and something subtle has happened.
The market has not changed the system.
We have.
The loss we defined before entering was acceptable when it was theoretical. Now that it is becoming real, we have decided the original rule no longer applies.
That is precisely the moment at which it must apply.
The initial stop exists so that we do not have to make an emotionally loaded decision while losing money.
Failure was anticipated.
Its terms were agreed before we arrived.
The Trailing Exit: Making Success Abnormal
Now we reach the harder rule.
Suppose the trade works.
Price moves in our favour. The position becomes profitable. Perhaps very profitable.
What should we do?
Human instinct has a wonderfully reassuring answer.
Take the money.
Lock it in.
Protect the gain.
Don’t let a winner turn into a loser.
It sounds prudent.
For an Outlier Hunter, it can be catastrophic.
Because the trade sitting in front of us may be the one we have been paying all those small losses to find.
We don’t know.
That is the problem.
The ten-R trade begins life looking like a one-R trade.
The thirty-R trade passes through ten R on its way to thirty.
At every point along the journey, the profit is available to be taken.
And at every point, taking it eliminates whatever might have come next.
This is why I do not think of the trailing exit as a device for protecting profits.
Its function is different.
It defines the conditions under which the system is finally willing to conclude:
Enough of the trend has changed. We leave now.
That is a very different instruction.
A genuine outlier does not travel in a straight line.
It surges.
Pauses.
Retraces.
Consolidates.
Looks finished.
Moves again.
Sometimes violently.
The uncomfortable truth is that we cannot know in real time which retracement is merely part of the journey and which one marks the end.
So once again we substitute a rule for knowledge we do not possess.
The trailing mechanism follows the position in the favourable direction while allowing sufficient room for ordinary variation around the trend.
Eventually price reaches it.
We exit.
Not at the top.
Not at the bottom.
After them.
That is not bad execution.
That is what a trailing exit is supposed to do.
“The trailing stop does not protect profits. It defines the conditions under which the programme concludes that the trend is over.”
You Have to Give Something Back
This may be the most difficult idea in the entire episode.
To capture a large trend, you must be willing to give some of it back.
There is no way around this.
If you demand the maximum profit from every position, you need to know where the maximum occurs.
You don’t.
Neither do I.
Neither does the system.
The only moment at which the exact top of a long trend becomes obvious is after price has moved away from it.
So the price of staying aboard is retracement.
Sometimes substantial retracement.
A position may show an enormous unrealised gain and then surrender a meaningful part of it before the trailing exit finally closes the trade.
That feels awful.
Nothing is being “lost” in the conventional sense, yet the emotional experience can feel exactly like loss.
Yesterday the account showed one number.
Today it shows a smaller one.
The mind immediately converts the difference into money we somehow owned and then surrendered.
This is where understanding convexity and actually living with convexity become two very different things.
Everyone likes the right tail in a backtest.
Living inside one is harder.
The great trade does not announce itself as the great trade.
It tests you.
It gives you profits and takes some back.
It creates reasons to leave.
It produces commentary explaining why the move has gone too far.
It makes prudence feel like intelligence.
And all the while the trailing rule may be saying:
Stay.
That is the harder discipline.
Taking a small loss hurts.
Watching a large unrealised profit shrink can be worse.
Because now we have something to protect.
The Most Expensive Sentence in Trend Following
There is a sentence I suspect has cost trend followers more money than almost any other:
“You can’t go broke taking a profit.”
Perhaps not.
But you can certainly destroy a convex strategy that way.
Imagine a programme designed around repeated small losses and rare enormous gains.
Now add a rule that takes every profit once it reaches two units.
What have we done?
We have left the losing side largely intact and amputated the winning side.
The distribution that justified the strategy has disappeared.
This is why profit targets sit so uneasily with Outlier Hunting.
A target says we know, in advance, how much success is enough.
But if the entire purpose of the strategy is to capture events whose magnitude we cannot know in advance, that assumption makes little sense.
Cocoa does not care that we have made five R.
Bonds do not stop repricing because our system has reached ten R.
A currency crisis does not consult our profit target.
The market decides how far the trend travels.
Our job is to remain attached for as long as our exit logic permits.
That is the kite.
We control the string.
We do not control the wind.
Non-Symmetry: Why Long and Short Need Not Be Mirror Images
There is another assumption worth questioning.
Why should a long trade and a short trade use identical rules?
It is aesthetically appealing.
Symmetry looks clean.
Markets have no obligation to be aesthetically pleasing.
In our programme, long and short parameters can be specified independently because the empirical behaviour encountered on each side need not be identical.
This is not a claim that every market always rises one way and falls another.
It is something more modest and more useful.
We do not impose symmetry where the evidence does not require it.
If robust testing across many markets and regimes suggests that different exit tolerances work better on the long and short sides, the architecture is free to reflect that.
The important word is robust.
We are not fitting separate parameters to every market until the historical equity curve looks beautiful.
That would be optimisation masquerading as sophistication.
We are asking whether broad structural differences persist sufficiently across markets and time to justify different treatment.
If they do, use them.
If they don’t, don’t.
Simplicity remains the default.
But simplicity does not require artificial symmetry.
The Exit Cannot Know Either
There is a pleasing symmetry between Episode 1 and Episode 2.
The entry does not know whether a trend is beginning.
The exit does not know whether a trend is ending.
Think about that.
At neither end of the trade does the system possess the information we would ideally like it to have.
Yet it operates anyway.
The entry says:
This condition has been met. Enter.
The exit eventually says:
This condition has been met. Leave.
Between those two events lies everything the market chose to give us.
Sometimes almost nothing.
Sometimes a loss.
Sometimes a modest gain.
And very occasionally, an outlier.
The system never knew which one it was holding while it was holding it.
It didn’t need to.
This is one of the deeper lessons of systematic trading.
Robustness does not come from eliminating uncertainty.
It comes from building machinery that can operate inside it.
Exit Logic Inside the Ensemble
Now put several systems together.
Different entry rules enter the same trend at different points.
Naturally, their exits need not occur together either.
One system may have been aboard for months.
Another may have entered much later.
Their trailing mechanisms may sit at different levels. A retracement can therefore remove one position while leaving another untouched.
The programme does not necessarily move from fully exposed to completely absent in a single decision.
Exposure can change progressively as individual systems reach their own exit conditions.
No committee decides this.
No portfolio manager looks at the combined position and says, “Let’s take half off.”
It emerges from the interaction of multiple independent rules.
This is another example of the principle we encountered in Episode 1.
The intelligence is in the architecture.
No individual exit knows the correct moment to leave.
No individual system knows how much of the ultimate trend it will capture.
But multiple simple mechanisms, operating independently across different markets and different manifestations of trend, create something richer than any one of them.
The ensemble does not need a master exit.
Its behaviour emerges from the exits beneath it.
Two Rules. Two Problems.
So we return to the kite.
The initial stop and the trailing exit are both pieces of string.
But they are doing different jobs.
The initial stop says:
If this doesn’t work, this is where I leave.
The trailing exit says:
If this does work, I have no idea how far it can go.
One makes failure survivable.
The other refuses to put a ceiling on success.
That combination is the engine of convexity.
Not prediction.
Not a high win rate.
Not finding perfect entries.
A bounded planned loss on one side and an open-ended possibility on the other.
Most trades will never exploit that possibility.
That doesn’t matter.
We only need a few that do.
And when one finally arrives, the hardest thing in the world may be to do absolutely nothing while a large unrealised profit becomes smaller.
But that is the bargain.
If we want the right tail, we have to give it room to exist.
We hold the string.
The market supplies the wind.
And we do not know how far the kite can fly.
That is not a problem for the system to solve.
That is the opportunity the system was built to capture.
READ DEEPER
→ Better Brakes, Faster Gains: Unlocking the True Power of Convexity in Outlier Hunting
→ The Feel of Convexity: Why Understanding Is Not the Same as Tolerance
→ The Convexity Edge: Why Embracing Uncertainty is the Key to Long-Term Success
→ The Hidden Costs of Certainty: Why the Best Trades Are Uncomfortable
Previous: System Anatomy 1: Entry Logic | Next: System Anatomy 3: Position Sizing
Richard Brennan writes on systematic trading, complex adaptive markets, and the philosophical foundations of trend following at atstradingsolutions.com. His books include The Fractals of Finance, Complex Adaptive Markets, Carved by Impossibility and The Aussie Turtles Trend Following Guide.
Want to explore why structure exists at all?
Carved by Impossibility: What Remains When Everything Else Is Eliminated
The book explores the architecture of constraint, emergence, and reality itself, and what it means for how we understand markets, life, and the universe.
Available now on Amazon in paperback, hardcover, and Kindle.
Want the theoretical foundation for why trend following works?
The Fractals of Finance: Determinism, Adaptation and the Geometry of Markets
The book explores the full architecture of feedback, fat tails, and fractal structure in financial markets, and what it means for how we trade, invest, and understand risk.
Available now on Amazon in paperback, hardcover, and Kindle.
Want a practical field manual for trading trends and capturing outliers?
The Aussie Turtles Trend Following Guide: A Field Manual for Hunting Outliers adapts the timeless principles of the original Turtle traders into a systematic, rules-based approach for modern markets. Co-authored with Adam Havryliv.
Available now on Amazon in paperback, hardcover, and Kindle.