Trading Journal Guide: Turn Trades Into Useful Evidence

A trading journal is not a diary of how you felt after winning or losing. Its primary job is to preserve the information needed to answer: what did I do, under what conditions, did I follow the plan, and what happened?

Educational information only. Historical journal results do not guarantee future performance.

Record the plan before the outcome

Post-trade memory is biased by the result. Capture important planned fields before or at entry:

• playbook/setup;
• instrument and contract;
• session;
• direction;
• context/regime;
• entry trigger;
• invalidation;
• initial stop;
• planned risk;
• intended management.

This makes rule adherence measurable.

Record execution separately

After the trade, store:

• actual entry/fills;
• actual exit/fills;
• quantity;
• fees/slippage;
• timestamps;
• partials;
• stop/target modifications;
• technical errors.

A strategy can be sound while execution is poor, or vice versa.

MFE and MAE

Maximum Favorable Excursion (MFE) describes the largest favorable move while the trade was open. Maximum Adverse Excursion (MAE) describes the largest adverse move.

These can help evaluate stop/target placement and management, but avoid changing rules merely to optimize past excursions.

Screenshots need context

A chart screenshot is useful when it preserves the timeframe, session, important levels, and state at decision time. A cropped image taken after the move can accidentally add hindsight.

If possible, keep pre-entry and post-exit views.

Tag market conditions

Useful tags may include:

• trend/balance;
• volatility regime;
• session;
• event proximity;
• location relative to VWAP/profile/reference levels;
• setup variant;
• long/short;
• rule adherence.

Keep the taxonomy small enough that tags are used consistently.

Journal mistakes, not just losses

A loss can be a perfectly executed strategy outcome. A winner can contain serious process errors.

Create separate categories for:

• compliant strategy loss;
• strategy win;
• execution mistake;
• risk violation;
• operational failure;
• ambiguous/unclassified.

This prevents P&L from becoming the only teacher.

Daily review

Daily review should be short and factual:

• Were all trades eligible?
• Did any rule get overridden?
• Was risk consistent?
• Were there operational problems?
• What should be investigated, not immediately changed?

Avoid redesigning the strategy after one session.

Weekly and monthly review

Aggregate by playbook, session, regime, direction, and adherence. Review expectancy, average win/loss, MFE/MAE, costs, drawdown, and error rates.

Look for enough observations before treating a pattern as evidence.

Use versioning

When a playbook changes, record the version/effective date. Otherwise old and new logic become mixed in one performance statistic.

Keep historical versions rather than rewriting old trades to match today's rules.

Journal qualitative decisions

For discretionary decisions, capture a concise reason and the evidence available at the time. Later, classify whether that discretion added value across a sample.

This turns “feel” into something that can eventually be evaluated.

AI-assisted journal review

AI can summarize notes, group recurring mistakes, retrieve similar examples, and suggest hypotheses. It should not invent missing fields or declare causality from a few observations.

Keep raw records accessible so summaries can be audited.

A minimum viable journal

If complexity prevents consistent journaling, begin with:

• setup;
• context;
• entry/exit;
• initial risk;
• result in R;
• adherence;
• one screenshot;
• one sentence explaining any deviation.

Expand only when a field will inform a decision.

The Journal Schema

Use stable fields: trade ID, strategy version, instrument/contract, timestamps, session, direction, setup, context, planned entry/invalidation, quantity, planned risk, actual fills, fees, realized P&L, R, MFE, MAE, adherence and incident tags. Stable schemas make aggregation possible.

Decision-Time Evidence

Capture the chart and reasoning before or at entry. A post-session screenshot contains information that did not exist when the decision was made. Keeping both pre-trade and post-trade views helps separate decision quality from hindsight.

A Sample Journal Record

A concise record might read: Playbook: VWAP reclaim v3; Context: balanced London session; Trigger: predefined reclaim; Initial risk: 1R; Outcome: -1R; Adherence: compliant; Note: no execution error. The loss is still useful evidence because the process was followed.

Separate Strategy and Behavior

Maintain distinct tags for strategy outcome, execution error, risk violation and operational incident. If a non-compliant trade wins, do not relabel it compliant. Otherwise profitable mistakes contaminate the strategy dataset.

Review by Cohort

Group trades by playbook, version, session, volatility/regime, long/short and adherence. Compare sample sizes with the metrics. A subgroup with three trades should not receive the same confidence as one with hundreds.

MFE and MAE Analysis

MFE and MAE can show whether winners regularly travel far beyond targets or whether trades routinely approach the stop first. Use distributions rather than optimizing to the single best historical threshold. The purpose is to generate hypotheses for testing.

Journal Expectancy

For each playbook calculate win rate, average win/loss, expectancy, costs and drawdown. Our What Makes a Trading Strategy Profitable? guide explains why these metrics must be interpreted together.

Review Cadence

Use immediate capture, a short end-of-session operational review, and slower weekly/monthly strategy review. Separating these timelines reduces the temptation to rewrite a playbook because of one emotional outcome.

Data Ownership and Export

Prefer systems that let you export or otherwise retain your own records. Back up important research. If using AI to summarize a journal, preserve the raw records so the summary can be checked against source evidence.

TensorAlgo Trade Review

TensorAlgo's current workflow automatically records supported Playbook trade activity and provides Trade Intelligence/review tools; plan limits can affect history and export availability. Check the current support documentation for authoritative feature details rather than hard-coding changing plan limits into an evergreen article.

External Reference

The CFTC Learn & Protect center is a primary source for futures-market education and risk. A journal should preserve enough detail to connect your own observations with authoritative product mechanics rather than relying on remembered assumptions.

Journal Database Design

Use one immutable trade ID and separate tables/fields for strategy version, fills, annotations and screenshots where possible. This prevents a later edit to a note from changing the underlying execution record.

Timestamps

Store entry intent, first fill, last fill, exit and important management actions with timezone-aware timestamps. Session analysis is unreliable when trades cannot be assigned consistently to the intended market day.

Costs

Record commissions and fees separately from gross P&L. For short-horizon strategies, also estimate slippage against the decision or expected price. Gross and net performance answer different questions.

R-Multiple Consistency

Define 1R from the initial planned risk and preserve it. If the stop is widened later, record that as a risk change rather than rewriting 1R. This makes risk violations visible.

Screenshots as Evidence

Capture pre-entry, management and post-exit views using consistent chart settings. Include enough context to see session/timeframe and relevant references. A screenshot without timestamp or scale can become decorative rather than analytical.

Error Taxonomy

Create a small stable set such as early entry, late entry, wrong size, stop violation, unplanned trade, missed valid setup and operational incident. Too many tags reduce consistency; too few hide useful patterns.

Monthly Research Questions

Ask which playbooks generated the most observations, which conditions changed expectancy, where execution cost increased, whether adherence improved and which apparent patterns still have too little data.

Avoiding P-Hacking the Journal

If you slice trades by enough tags, some subgroup will look exceptional by chance. Treat discoveries as hypotheses and test them on later trades rather than immediately changing the playbook.

Privacy and Security

Journals can contain financial and behavioral information. Use secure authentication, backups and appropriate sharing controls. Remove credentials or sensitive account identifiers from screenshots before sharing publicly.

AI Review Workflow

Ask AI to summarize only sourced journal fields, link claims back to examples and separate observations from hypotheses. A useful output might say '12 of 18 late entries occurred after X tag' rather than inventing a psychological cause.

A Journal Schema You Can Actually Analyze

A useful journal separates fields that describe the decision, execution, market context, and outcome. Mixing everything into free-form notes makes later analysis difficult.

| Group | Example fields | | --- | --- | | Identity | timestamp, market, contract, session, playbook/version | | Decision | setup eligibility, trigger, invalidation, vetoes, confidence category | | Risk | planned risk, size, stop distance, remaining daily capacity | | Execution | order type, intended price, fills, slippage, fees | | Context | regime, volatility, event window, key references | | Outcome | realized R/P&L, MFE, MAE, exit reason | | Process | adherence, mistake category, screenshot, technical incident |

Use stable categories. If “late entry” means something different every week, the field cannot support a reliable cohort analysis.

The Rule-Based Trading guide helps turn vague setup descriptions into consistent fields, while the Trading Playbook guide provides the named setup structure a journal can reference.

Preserve Decision-Time Evidence

The most valuable screenshot is often the one captured when the decision was made. A chart reconstructed after the outcome is known invites hindsight. Store the information that was actually available: active levels, relevant context, planned invalidation, intended management and the reason the setup qualified.

For automated or AI-assisted workflows, also record model/version, feature timestamp and any recommendation or classification that influenced the process. This creates provenance and makes later debugging possible.

Review by Cohort, Not by Anecdote

Group trades by playbook, version, session, regime and adherence. Compare compliant and non-compliant trades separately. A strategy problem and an execution-discipline problem require different responses.

MFE and MAE can reveal whether exits or stops deserve investigation, but they should not be used to optimize every trade retrospectively. The Backtesting Guide explains why repeated tuning against observed outcomes can overfit noise, and the Forward Testing Guide shows how a proposed change can be evaluated on new observations.

Journal the Missing Trades Too

Where practical, record valid setups that were missed and invalid trades that were taken. Otherwise the journal only contains executed trades and cannot answer whether attention, alerts or qualification rules are creating selection bias. The Decision Fatigue guide is useful when missed or impulsive decisions cluster after long sessions or repeated choices.

Data Ownership and Privacy

Prefer a journal that can export your records in a usable format. Back up important evidence and treat broker/account identifiers, credentials and personal financial information as sensitive. If screenshots are shared with external services, review what they contain first.

A journal is not valuable because it is detailed. It is valuable because the data is consistent enough to compare decisions over time and specific enough to tell strategy, execution, behavior and operational failures apart.

Weekly Review Template

A weekly review should answer a fixed set of questions before looking for new patterns. How many valid setups occurred? How many were taken or missed? Which trades followed the playbook? What were expectancy, average win/loss, MFE/MAE and costs by playbook? Did any operational incidents affect results? Were mistakes concentrated by time of day, session length or recent outcome?

Then separate actionable process fixes from research hypotheses. Forgetting to record a screenshot can be fixed immediately. Suspecting that a stop should be wider requires a historical test and new forward evidence.

Avoid p-hacking the journal

A rich journal makes it easy to slice results into dozens of tiny groups until one looks impressive. Treat exploratory findings as hypotheses. If “Tuesday afternoon trades after two losses” appears unusually weak in a small sample, do not instantly ban them; define the rule and test it on additional data.

Keep taxonomies stable

When adding a new mistake or regime category, document its definition and start date. Reclassifying old trades after seeing outcomes can create hindsight bias. If historical recoding is necessary, use an explicit rule and apply it consistently to the full sample.

Journal system quality

Track missing fields and data quality. A metric based on 40% of trades is different from one based on nearly complete capture. Periodically audit whether timestamps, fees, screenshots and playbook versions are actually present.

The goal is a dataset that can survive skeptical review. A journal becomes valuable when another competent reviewer can understand what was known, what was planned, what happened and how the trade was classified without relying on the trader's memory.

From Journal to Research Queue

A journal should produce a small, prioritized research queue rather than endless observations. Require a minimum description for every proposed investigation: observed pattern, sample affected, plausible mechanism, expected change and the test that could reject it.

Rank hypotheses by potential impact and evidence quality. A repeated execution error across many trades deserves attention before a speculative pattern seen twice. This keeps review from becoming a search for novelty.

When research produces a rule change, record the new version and effective date in the journal. That closes the loop between observation, testing and production behavior, and prevents later analysis from mixing trades generated by materially different rules.

Completion Matters

A smaller journal with consistently completed fields is more useful than a sophisticated schema filled only after memorable trades. Track completion rate and simplify fields that are routinely skipped. Data quality is part of the trading process because every later conclusion inherits the weaknesses of the records used to produce it.

Frequently Asked Questions

What should a trading journal record?

Capture the playbook and version, decision-time context, entry and exit, risk, fills and costs, management actions, adherence, MFE/MAE and notes needed to reconstruct the decision.

What are MFE and MAE?

Maximum Favorable Excursion and Maximum Adverse Excursion describe how far a trade moved in the favorable and adverse direction while it was open.

Are screenshots enough for a trading journal?

No. Screenshots are useful evidence, but they need timestamps, rule context and structured fields so trades can be compared across a sample.

How often should a trading journal be reviewed?

Use a cadence that separates immediate capture from batch analysis—for example, a short daily check plus deeper weekly or monthly reviews when enough observations accumulate.

Final Takeaway

A useful trading journal preserves what was known and intended before hindsight. Structured decision-time context, execution, risk, adherence and excursion data turn individual trades into evidence that can improve a playbook over time.