Mistakes & Risk

If the strategy needs too much parameter precision, too many filters, or too much explanation after the fact, the edge is probably not durable enough for capital.

When position size outruns evidence, ordinary drawdowns start forcing emotional decisions, and the system becomes unstable even if the signal logic stays sound.

Not watching health, execution quality, or regime change is still an active decision. The absence of intervention can itself be the mistake.

A strategy that no longer earns desk space should be cut cleanly. Endless rescue attempts usually waste more capital and time than the original loss.

Ask what market condition would make the strategy wrong, how much damage is acceptable, and whether the live environment can surface those signals quickly.

Define the drawdown, technical failure, or behavior drift that ends the strategy's right to keep trading.

Even good systems fail. Capital allocation should assume uncertainty rather than reward confidence.

When evidence says the system no longer fits its environment or its risk budget, close it. Replacement beats denial.

• You cannot explain why the strategy still deserves capital.
• You keep widening tolerances because the loss feels temporary.
• You are changing rules faster than you can validate them.

• No bot gets capital without explicit stop conditions.
• No size increase happens without evidence from review and validation.
• No weak strategy stays alive because of attachment or sunk cost.

• Know how the strategy fails.
• Know how the operator intervenes.
• Know what level of damage still counts as acceptable variance.