A useful journal lets you reconstruct a decision and reconcile it with the account history. Keep numerical records and short notes made at the relevant time. Choose fields from the questions you need to answer; neither a particular column count nor regular review guarantees better returns.

Separate a trading idea from its executions

One position can have several opening fills and partial exits. Preserve the raw operation history and give each planned idea its own ID. Link the fills, exits and costs to that ID. Two exits do not automatically become two independent strategy observations.

Document your grouping rule. MT5’s report defines a trade for its weekly count as an operation registering a financial result, including a partial exit or reversal. Your idea count can therefore differ from the terminal count while the cash totals still agree. Investigate unexplained differences rather than force the same number.

Record the account, instrument, direction and rule version at the decision; entry conditions, initial stop and planned risk before the order; execution time with timezone, price, quantity and identifiers from the history; and changes when they happen. Keep a separate net result, currency, risk denominator and review note. The article on keeping a trader’s journal develops the recording habit.

Add costs that already carry a minus sign

A fictional export has separate fields P/L: +50 USD, commission: −3 USD, swap: −2 USD. They contain signed cash changes and P/L excludes these costs. That convention matters more than the column name.

Negative costs reduce the result when addedP/L +50 USD, commission −3, swap −2. A waterfall falls 50 to 47 to 45. Signed amounts sum to 45. Subtracting the negative costs incorrectly gives 55.Signed amounts in USD+50−3−245P/LFeeSwapNet50 + (−3) + (−2) = 45 USDWrong: 50 − (−3) − (−2) = 55
Common scale6px/USD; costs reduce50→47→45. An import model, not a broker-report screenshot.

Net equals 50 + (−3) + (−2) = 45 USD. “P/L minus commission minus swap” gives 55 after inserting already signed amounts. The 10 USD error doubles the 5 USD costs. Define whether an imported field is a positive cost amount, a signed change or a net result, and reconcile it with raw operations. Swap may be positive or negative; do not automatically replace it with its absolute value.

Assumptions: One currency, three separate components, no other fees. P/L excludes costs; commission and swap are negative. Positive cost magnitudes or already-net P/L need a different formula. Official reference: MQL5 — deal properties.

Keep initial risk and reconcile net result

Own hypothetical EUR/USD model: USD account, 100,000 EUR per lot, pip = 0.0001, 0.20 lot and a planned 25-pip stop. The pip value for one lot is 10 USD, so the initial planned price risk is 0.20 × 10 × 25 = 50 USD. Here that denominator deliberately excludes fees. It is a planned reference, not a guaranteed maximum loss.

Close half at +30 pips and the other half at +10 pips: 0.10 × 10 × 30 = 30 USD and 0.10 × 10 × 10 = 10 USD. With total costs of 3 USD, net is 37 USD and the idea result is 37/50 = 0.74R. Keep the original 50 USD denominator after a stop move. Additional position size requires a declared rule for the whole idea’s risk.

Define each import field before using a formula. The adjacent diagram uses a separate signed-cost model: gross P/L +50, commission −3 and swap −2 give 45 USD when added. A net field already including those costs must not have them charged again. Cash deposits, withdrawals and non-trade charges also need their own categories when reconciling the account.

For MAE/MFE, state price side, unit, sign and the observed lifetime. A separate long-position model enters at Ask 1.1002. While it is open, recorded Bid prices are 1.1000, 1.0994, 1.1008 and the exit 1.1006. With pip = 0.0001, the largest observed adverse distance from entry is 8 pips and favourable distance 6 pips. A later Bid 1.1020 after exit is excluded. These distances do not establish available money after fees or an executable extreme.

Mark missing intratrade observations as unavailable, not zero. A bar extending before entry or after exit may include irrelevant extremes, and changing quantity needs more than price maxima to reconstruct monetary exposure. See MAE/MFE analysis and trading statistics. A mean calculated only over winners does not by itself identify a better stop for the full strategy.

Write observations at the time of the decision

A note such as “I sent the order before the required bar closed” can be checked against timestamps and the rule version. A profitable outcome does not make that action compliant. A larger-than-1R loss can arise from costs or adverse execution without proving a discipline failure.

A consistently defined self-rating is optional. Record it before entry if you intend to study it; do not reconstruct an old feeling from the later result. Notes preserve observations and stated reasons, not proof that a feeling caused a loss. Keep corrections visible and avoid changing a setup label just to remove a losing trade from its original group.

Quick check. Select a small known set including a loss, partial exits and financing. Can every idea be linked to the raw operations, original risk and net total? List the missing fields and unexplained amounts. The check does not require a universal seven-of-ten threshold or a diagnosis of self-sabotage.

Review decisions, including the ones that went well

Start with complete records, duplicate detection and cost reconciliation. Review rule deviations, unusual fills and larger losses, but include correct decisions and profitable mistakes. Filtering only losses below −1R can miss repeated smaller errors.

Keep simulated outcomes of omitted signals separate from executed money. Define eligibility and fills before calculating them. A time-of-day association can be a hypothesis; it does not isolate a cause. Record sample size, instrument, position size and market conditions before proposing a new rule version.

Finish with a specific next task: repair an import, clarify an ambiguous rule or test a recorded hypothesis on unused observations. A review can also conclude that no change is supported yet. Set a schedule you can maintain without requiring a particular duration or a new strategy change each week.

Choose software by the records it can preserve

A spreadsheet can support formulas, charts and imports; Excel’s Power Query offers data import and transformation with platform/version limits. Check your licence and available connectors. See Excel templates for the next formula checks.

For a dedicated journal, verify your broker import, partial-close grouping, costs, currencies, export access and current plan on the provider’s own site. Test it against known operations before paying for a dashboard. An importer cannot recover a thought or original stop absent from its source. The trader’s workshop covers related tools.

Two monthly reports can answer different questions

A separate hypothetical idea has two exits: +30 USD on 31 January and +10 USD on 1 February, each with a 1 USD fee. Total net is 38 USD. Grouping by each realization gives January 29 and February 9 USD. Assigning the entire completed idea to its last-exit month gives January 0 and February 38 USD.

The same 38 USD under two declared grouping rules
Report purposeJanuaryFebruaryTotal
Realized operations by their date29 USD9 USD38 USD
Completed idea by last-exit month0 USD38 USD38 USD

The second report alone cannot explain January’s realized +29 USD in account history. The two monthly pieces in the first report do not become two independent ideas. Preserve the idea ID and dated costs, label the report’s question and reconcile the full history. Neither view substitutes for equity including an open remainder or a jurisdiction’s tax records.

Start with a sample you can verify

  1. Define the records. Separate raw operations, idea IDs, initial planned risk, costs and contemporaneous notes. Store different numerical values in separate fields.
  2. Reconcile known examples. Include a partial exit, a loss and financing. State sign and currency conventions; mark unavailable history instead of guessing.
  3. Review process and results separately. Include compliant decisions and profitable deviations. Record a specific follow-up question without requiring a diagnosis or a rule change.
  4. Keep rule versions visible. Later comparisons need sample sizes and comparable definitions. A before/after increase in mean R alone does not establish that journaling caused it.