Build a journal that tells the truth

Write before the outcome teaches you a story

After a profitable trade, the original decision often feels more obvious than it did beforehand. After a loss, it is tempting to claim that a warning was always visible. A timestamped journal preserves what was actually known and expected. It is not a motivational diary; it is a record that lets a later reviewer distinguish evidence from reconstructed memory.

Before a simulated trade, record the instrument, product type, observation time, hypothesis, planned entry and exit rules, position size, cost estimate and reasons not to proceed. Include the event calendar and any unresolved uncertainty. A short clear sentence is more useful than several paragraphs of confident language that never state what would make the idea wrong.

Use separate timestamps for the decision note, order submission, execution and later review. A field entered after the result is known should not masquerade as a pre-trade expectation. Corrections are sometimes necessary, but preserve the original value and append the correction with a reason. This makes the journal useful even when the original entry was incomplete: an honest missing field reveals more than a polished reconstruction that hides what was actually unknown.

Separate execution from outcome

A profitable result can follow a poor decision, and a well-specified process can produce a loss. Evaluate whether the decision followed the recorded rules before looking at the money. Then evaluate whether those rules themselves deserve continued testing. Otherwise, favorable randomness rewards improvisation while ordinary losses encourage unnecessary changes.

For execution, record actual or simulated fills, spreads, fees, timing and deviations. For behavior, record observable actions: moved the exit after price declined, increased size after a loss, or entered without the required condition. Avoid vague labels such as bad discipline when a specific action can be measured. The aim is to identify a repeatable failure point, not to assign moral blame.

Recognize common distortions

Confirmation bias can make supporting evidence easier to notice than contradictory evidence. Anchoring can give an arbitrary purchase price too much influence over a later decision. Overconfidence can grow after a lucky run. These tendencies are not diagnoses of a particular person; they are useful questions to ask when reviewing a decision trail.

A journal can introduce a small countermeasure: require one disconfirming fact before entry, compare the decision with a predefined checklist, and note whether the same action would make sense without knowing the purchase price. None guarantees good decisions. Their purpose is to make reasoning more inspectable and to create friction before an impulsive action.

Review in groups rather than chasing every outcome

Changing a rule after each trade makes it difficult to know what was tested. Review a defined batch of observations and separate strategy changes from execution fixes. Correcting a unit error is different from changing the entry threshold to rescue disappointing performance. Keep versions so that results from one rule set are not silently combined with another.

The review should include skipped trades and reasons for skipping. Otherwise the journal can become a selective collection of memorable actions. Track time spent as well as money: a process that consumes hours has an opportunity cost even in simulation. Protect account identifiers and sensitive information if sharing screenshots or journal files with another person.

A weekly summary might count how many entries had complete notes before submission, how many fills exceeded the cost estimate, and how often an order was changed without a recorded reason. These process measures are less glamorous than a profit chart, but they directly address whether the exercise was conducted as designed. Keep conclusions proportionate: a handful of entries can identify a mistake without establishing a durable behavioral pattern.

Build a review that can distinguish three different problems

A review should distinguish hypothesis failure, implementation failure and recordkeeping failure. The hypothesis may have followed its rules yet produced an unfavorable sample. Implementation may have deviated through delayed orders, wrong quantities or unrealistic fills. Recordkeeping may be too incomplete to tell which occurred. These are different findings. Rewriting the signal cannot fix a unit-conversion error, and tidying a journal cannot create a profitable relationship that the data do not support.

Define process measures with visible denominators. If 18 of 20 executed paper trades had complete notes before entry, completion was 90% for executed trades. That does not establish that every eligible signal was recorded. Keep a separate count of all signals, including those rejected by the rules, skipped for operational reasons or overlooked entirely. Otherwise a tidy set of trade entries can hide a selective decision process.

If outcomes are expressed in R units, define R from the initial recorded planning loss and do not change it after the result. A trade with an initial $100 planning loss that actually loses $140 has an outcome of −1.4R. Redefining R as the realized loss would make every losing trade appear to lose exactly one unit and conceal gaps or execution errors. Keep dollar results, initial planned risk and the reason for any difference together.

Counterfactual notes can be useful, but label them as simulations. “Had the original exit remained unchanged, the modeled result would have been…” depends on assumptions about price sequence and fills, not just a visible chart. Do not add that imagined result to actual or paper-account performance. Use it to identify a question for the next version of the exercise, and decide in advance what evidence would show that the proposed change helped rather than merely fitting the last disappointment.