Everyone agrees checklists are wise until a checklist blocks a trade they love. Then the document becomes “bureaucracy,” “overfitting to fear,” or “not how experienced people work.” Ego is the main failure mode of process tools. Building checklists that survive ego means designing for the day you feel certain, not the day you feel humble.
Short beats comprehensive
A thirty-page checklist will not be used under time pressure. A one-page list of non-negotiables might. I favour items that gate capital: thesis written, invalidation clear, size within limits, liquidity exit plausible, correlations checked, operational path confirmed. If an item cannot stop a trade, it may belong in a longer manual, not on the gate list.
Checklists should be specific enough to be auditable. “Consider risk” is not a checklist item. “Confirm gross exposure remains under X after entry” is.
Make bypass visible
People will bypass process. The question is whether bypass is visible. Require a written exception with an expiry date. Exceptions that live forever become the new culture. Calm markets are when silent exceptions multiply because “nothing bad happened.” That is how limits die.
Identity-safe framing
Frame the checklist as professional athletic gear, not as a moral judgment. Pilots and surgeons use checklists because the domain punishes improvisation, not because they lack skill. Investment teams can borrow that framing. The goal is fewer unforced errors, not the humiliation of experts.
Digital assets and AI workflows need the same spine
New asset classes and new tools tempt people to declare old checklists obsolete. Sometimes details change—custody steps, oracle checks, data-leakage tests. The spine should not: write the idea, size the risk, name the failure, confirm you can exit, record the decision. Novelty is not a waiver.
For AI decision support, include a human confirmation that model output was interpreted within its training and monitoring bounds. “The model liked it” is not a thesis.
Maintenance is part of the design
Design for the confident day
Most checklists are written after mistakes, when humility is high. They must be usable when humility is low. That means fewer items, clearer verbs, and no moralising language that invites rebellion. “Confirm invalidation price is written in the order ticket notes” survives better than “remember to be careful.”
Pair checklists with tooling. If the order management system can block a ticket missing a size field or a risk check, ego has less room to manoeuvre. Where tooling cannot block, require a second person for material size—even if the second person is a peer rotating the sceptical seat.
Post-trade honesty
Checklists before entry are incomplete without after-action review. Did we skip steps? Did an exception help or merely soothe? Close the loop within a week while memory is accurate. Without closure, the checklist becomes a pre-trade ritual unrelated to learning.
Refresh training when new joiners arrive. Culture transmits faster than wikis. If seniors joke about “the silly list,” the list is already dead. Seniors should model completion in public on ordinary trades, not only on ceremonial ones.
Ego wants freedom. Markets invoice freedom without rails. Build rails that skilled people are willing to use because the rails make winning less random and losing less existential.
Review the checklist after incidents and after boring quarters. Remove items that never change decisions. Add items that would have prevented a real mistake. Keep it alive without letting it bloat. A dead checklist is decoration. An obese checklist is sabotage.
Checklists for exits as well as entries
Ego interferes with exits too—especially when a thesis becomes identity. Exit checklists should ask whether invalidation hit, whether size still matches liquidity, and whether you are holding for vindication. Vindication is not a mandate.
Publish anonymised examples of checklist catches that prevented pain. Positive stories help ego accept the tool. Fear alone does not sustain habits.
Version control for process
Treat checklists like code: version them, note why an item was added, and retire items that never change decisions. Stale items breed contempt. Contempt kills compliance. A short living list beats a long museum of past fears.
In digital-asset and AI workflows, add specialised lines only when they gate capital—custody confirmation, leakage check, envelope confirmation—then keep the universal spine intact. Ego argues that specialists should be exempt. Markets disagree.
Make completion visible in ordinary weeks. Rituals that appear only after disasters feel like punishment. Rituals used on boring Tuesdays feel like craft.
If your process only works when everyone is already careful, it is not a process. It is a compliment. Build tools that still work when confidence is high and memory is selective. That is the only time process earns its keep.