For a long time, I believed entries were the hard part of trading.
Find the trend.
Define the regime.
Build the signal model.
Press the button.
But after enough trades, it became obvious that entries are easy compared to exits.
Trailing stops, in particular, exposed how fragile discretionary trading really is.
The Illusion of Simple Stops
On paper, trailing stops sound trivial:
Move the stop up as the trade moves in your favor.
In practice, every stop decision is a trade in itself.
Tighten too early and you miss the trend.
Tighten too late and you give back open profit.
Tighten inconsistently and your equity curve becomes noise.
There is no single correct answer in real time.
There is only a distribution of outcomes.
Why Entries Feel Easier
Entries are discrete decisions:
- Long or short
- Now or later
- Yes or no
They are binary and explainable.
Exits are continuous decisions:
- Tighten a little
- Tighten a lot
- Hold
- Move to breakeven
- Trail structure
- Trail EMA
- Do nothing
Exits require judgment on every bar while PnL is flashing in your face.
That cognitive load matters.
The Emotional Tax of Exits
I noticed something in my own trading:
I could explain why I entered a trade.
I struggled to explain why I exited.
Exits were influenced by:
- Fear of giving back profits
- Frustration after a pullback
- Hope that a trade would “come back”
- Regret after getting stopped out too early
In other words, exits were emotional, not engineered.
Machine A and Machine B
To remove myself from exit decisions, I split my trading system into two conceptual machines.
Machine A decides what to trade.
Machine B decides how to manage risk after entry.
Machine A handles signals, regimes, and entries.
Machine B handles trailing stops and exits.
Separating the two forced me to treat exits as an engineering problem, not a psychological one.
Why Exits Shape the Equity Curve
Two traders can take the same entries and end up with radically different results.
The difference is almost always exits.
Trailing stops determine:
- Average win size
- Tail risk
- Drawdowns
- Equity curve smoothness
- Psychological survivability
Entries find opportunity.
Exits define performance.
A Working Hypothesis
I’m testing a simple idea:
Thesis: If trailing stops are systematic, consistent, and adaptive, the equity curve will take care of itself.
That hypothesis led me to build a machine-learned trailing stop engine—what I call Machine B.
In the next post, I’ll explain how I framed trailing stops as a machine learning problem and how I defined “good” and “bad” stop decisions without cheating with future data.
Related: I Had Rules for Entries and Feelings for Exits →
Next: Random Forests for Trailing Stops: Labels Without Lookahead Bias →