If You Don’t Know the Worst Case, You Don’t Know the Size

Your model’s “risk” is usually a stop distance. The market’s risk includes gaps, slippage, spread explosions, margin changes, and outages.
Worst-case is not the same as “unlikely”. It is “what could happen when conditions are bad”.

A strategy can survive normal days and still die from one tail event.
If you size as if the stop price is guaranteed, you are underestimating the true distribution of outcomes.

The job is not to predict the tail. The job is to survive it.

Process upgrade: turn this truth into a rule you can actually enforce. If the rule can’t be enforced, it will be broken under stress.

• Write the trigger condition in observable terms.
• Write the action in one sentence.
• Write the penalty for breaking it.

Size based on stop → tail event (gap/slip/outage) → loss exceeds plan → margin call → forced liquidation → strategy abandonment.

How it kills accounts:

1. The rule exists only in your head.
2. Stress arrives and you improvise.
3. Improvisation becomes inconsistency.
4. Inconsistency becomes random results.
5. Random results become a slow bleed.

Write a worst-case scenario per instrument: gap size + spread + slippage + margin change.
Size so that worst-case is painful but survivable.
Keep margin buffers that assume conditions get worse before they get better.

Rule that survives:

• Write the trigger in observable terms.
• Write the action in one sentence.
• Write the penalty for breaking it.

You size a position so the stop equals 1R. A weekend gap skips the stop and the loss becomes 4R. Your plan was not wrong; your assumptions were incomplete.

Tell: if your rule isn’t enforceable, it’s a suggestion. Suggestions don’t survive stress.

What to do instead

Build your “worst case” from concrete components: gap risk, spread, and slippage. Then add operational shocks: margin changes and platform outages.

This is the same survival logic as The one bad day problem.

Related

Broker reality: Margin requirements may change without notice and No guarantee of continuous service.

Field checklist

• Write the rule in observable terms: if X, then Y.
• Remove choices under stress. Choices become rationalizations.
• Track the precursor: what state were you in before the mistake?
• Make deviations costly (size down, pause, review).
• Turn lessons into gates, not notes.