Trading journal — records, reconciliation and decision review

A trading journal connects the execution record with the original plan and the evidence used to make a decision. It can expose missing records, procedural breaches and outcomes of compliant trades. It does not replace the broker statement or automatically explain why a strategy makes or loses money.

Define what one journal record means

Decide whether a row represents one fill, a closed position or a trading idea with several entries. Partial fills and exits can produce several records for one idea. Retain account and execution identifiers so that regrouping neither duplicates a trade nor loses a commission. Keep the original history alongside your notes.

Possible review questions include how compliant and noncompliant entries differ, what happened after a loss, and whether results vary across documented setups. An entry following a loss is not automatically trading to get even. Specify the observable breach, and include ordinary outcomes rather than selecting only memorable wins or losses.

A subgroup difference suggests a hypothesis. It does not prove that the recorded feeling caused the outcome: position sizes, market conditions and strategy rules may differ too. A journal may show that there is too little comparable information to decide.

Ten groups of information to preserve

These are information groups, not a compulsory column count. An exit based on time or a moving rule need not have a fixed take profit. Preserve the original stop alongside later changes. Do not infer an emotion such as fear of missing the move from the final result alone.

Hypothetical EUR/USD buy: entry fill 1.0852, initial stop 1.0822 and target 1.0902. Assuming a pip of 0.0001, the distances are 30 and 50 pips. The planned reward:risk ratio is 50/30 ≈ 1.67:1; in risk:reward order it is 1:1.67. That is a plan before additional costs, not a realised result.

Assume 100,000 EUR per lot and a USD account. A position of 0.10 lot is 10,000 EUR, with a pip value of USD 1. Its stop-level price loss is USD 30 and target price gain USD 50. Calling it “0.1 of a mini lot” would mean a different quantity under that convention. Actual contract specifications and fills matter; an ordinary stop does not guarantee its requested price.

Reconcile the spreadsheet with the source

Keep source results and charges with their currency and sign convention. If charges are separate signed entries, add them with those signs. If a cost is recorded as a positive debit, subtract it. Do not apply one formula blindly to every export or subtract spread again when it is already reflected in entry and exit fills.

In our USD example, price result +25, commission −2 and swap +1 give net +24. If the source already supplies net +24, subtracting commission again would produce the wrong result. Include charges on both sides and separately booked fees without attributing one charge to several positions.

Define initial R before evaluation. With R = USD 30, a closed net +24 is +0.8R; net −33 is −1.1R. Document whether planned costs enter that denominator, and retain it after closure. Average winning pips divided by a planned stop cannot substitute for actual net results in R.

Deposits and withdrawals are not trading gains or losses. Closed results do not automatically include the changing value of open positions. Convert genuinely different source currencies to one declared reporting currency and valuation time; do not repeat a conversion the source already made. An educational journal is not automatically a tax calculation.

Use a supported table import or a provider export. Check decimal separators, time zones, the covered period, counts and totals against the original. Preserve the downloaded file and its retrieval time. Choose an application by import accuracy, export options, privacy and cost, without an obligatory three-month wait.

Test repeated imports too. Our fictitious import A contains execution IDs 101, 102, 103 with net +12, −8, +4 USD; import B contains 103, 104, 105 with +4, −5, +3 USD. Concatenating produces six rows and +10 USD. Five distinct executions produce +6 USD. The +4 difference is the duplicated execution, not extra trading profit. Key by account and execution ID, and flag conflicting values rather than silently choosing one.

Four questions for a review

Is the dataset complete and reconciled?

Check all qualifying trades, duplicate IDs, fees and unresolved records first. The largest wins and losses can guide investigation, but a selected extreme cannot represent the whole group.

Were the stated rules followed?

Assess observable conditions separately from profit. A profitable breach remains a breach; a compliant loss alone does not demonstrate an execution mistake. Keep unknown cases visible rather than silently treating them as compliant.

What does a grouping actually show?

The total is 250 − 180 − 320 − 40 = −290 USD. Haste and recovery labels cover 13 trades with −500 USD together. These assigned numbers are an arithmetic exercise, not observed trader statistics or evidence that emotions caused losses. Check position size, method, market conditions and actual breaches before interpreting the association.

What hypothesis or correction needs testing?

A confirmed duplicate requires an import correction. A subgroup difference may instead require further observations. Predeclare the hypothesis and evaluation rule; a review does not require a fresh trading restriction every week.

Avoid reconstructing an invented plan

Record the plan before the decision; add execution commentary when it does not interfere with managing risk. Automate facts where possible. A chart screenshot should identify its time, symbol and interval. Mark a later screenshot with future prices as review material.

Recover historical parameters from the execution record. Mark an uncertain feeling as missing and a later recollection as retrospective. Do not present a plan reconstructed from the outcome as a contemporaneous entry rationale.

Build and test a manageable journal

1. Choose the record unit and preserve the source. Import the supported history format, identify duplicates and reconcile costs and currency before interpreting groups.
2. Use a short context template. Record relevant cues without obstructing account supervision. Unknown information remains unknown.
3. Budget the review explicitly. A hypothetical 30 entries at two minutes each plus four reviews at 30 minutes is 180 minutes, or three hours. That is a budget calculation, not a time promise.
4. Evaluate a specific correction or hypothesis. Connect it to an observable test, as in building a system of trading discipline. Continued logging does not guarantee automatic error recognition after three months or profitable trading.