At 3pm, four hours into a choppy session and $180 down, the person deciding your next trade is not the calm student who read the risk books. It is a tired, mildly angry version of you with a shorter time horizon and a looser definition of 'setup'. Professionals solve this the way constitutional lawyers do: decide everything decidable in advance, write it down, and demote the in-session self from policymaker to clerk. The clerk executes. The clerk does not legislate.
Why written beats known
A rule that lives in your head is a suggestion, renegotiated in real time by whoever you happen to be at the moment of temptation. Written rules create a binary that survives stress: any trade either complied or it did not, and no narrative changes which. The gap is measurable in dollars. 'About 1 percent' under pressure quietly becomes $250 on a $10,000 account; a written 'exactly $100, computed before entry' stays $100. Writing also makes auditing possible: a journal can only test a rule that exists in a fixed form. One page, printed, taped where you trade. If it needs a second page, it is documentation, not a constitution.
A starter constitution
- Risk per trade is 1% of current equity, computed before entry, every entry.
- The stop goes in with the order, at a structure level, and never widens for any reason.
- Daily loss limit: 3% or three full losses, whichever comes first; the platform closes for the day when it's hit.
- Correlated positions count as one trade; aggregate exposure to a single theme never exceeds 2%.
- No entries within 10 minutes of a red-flag release on the traded instrument, and open positions get the flatten-reduce-hold review before it.
- Flat before the weekend close, every week, no exceptions.
- Two consecutive stop-outs on one setup retires that setup for the day.
- Below minus 10% from equity high, risk halves to 0.5% until a new high prints.
- No size increase within 24 hours of a losing day.
- Every trade is journaled the same day with a screenshot, the plan, and a grade; an unjournaled trade counts as a broken rule.
Steal this list wholesale or edit it, but keep two properties. Every rule contains a number, because numbers are checkable and adjectives are not. And every rule is about behavior you control, never about outcomes you don't: 'make $200 a day' is not a rule, it's a wish with a deadline.
Make breaking a rule expensive
A rule without a consequence is decoration. Attach an immediate, automatic cost: any broken rule ends the session on the spot, gets logged in red in the journal, and three logged breaks in a month cut size by half for the following month. The consequence must not care whether the broken-rule trade won. A $300 winner taken off-plan is the most expensive trade you can make, because it teaches the 3pm clerk that legislation is optional when it pays. Some traders add an accountability layer, sending the page to a trading friend who sees the break log weekly. Embarrassment is an underrated risk management tool.
Amend on weekends, never mid-fight
A constitution that cannot change calcifies; one that changes mid-session is worthless. The compromise: amendments happen only on weekends, in writing, with a journal-based reason attached. If rule seven keeps retiring setups that then work, the journal will show it across a real sample, and a weekend review can loosen it deliberately. Audit the whole page monthly the same way: total what each rule saved and cost, the exact exercise you'd run on a break even habit. Rules that earn nothing get retired, because a list that grows past roughly ten stops being read. The document is alive. The trading day is just never the moment it is allowed to change.