Eight preset questions can turn a large trade log into a focused review without turning every month into a research project. “One-click” means the filters and definitions are saved in advance; “10-minute” is a review target once the evidence is already imported, mapped, and reconciled—not a promise for a raw or incomplete journal.
The presets cover setup, session, weekday, post-loss timing, high-frequency days, direction, instrument, and an evidence report card. Each produces an observation plus a next test. None can identify motive, prove causality, or guarantee that removing a weak-looking group will improve future performance.
Before Preset 1: Pass the Evidence Gate
A fast diagnostic is only as good as its shared scope. Record the accounts, strategy versions, instruments, date range, timezone, currency conversion, cost coverage, and included trade states. Then check duplicates, missing timestamps, unmapped setups, unknown directions, and incomplete P&L.
| Gate | Pass condition | If it fails |
|---|---|---|
| Identity | Every row belongs to the intended account and source | Reconcile or exclude the entire unresolved source with disclosure |
| Outcome | Closed-result coverage and available costs are known | Label totals partial; do not rank by money alone |
| Labels | Setup, direction, symbol, and session coverage are visible | Show Unknown as a group instead of dropping it |
| Time | Entry timestamps use one declared analysis zone | Do not compare session, weekday, or post-loss timing |
| Comparability | Capital, risk policy, and strategy versions are documented | Split incompatible periods before interpreting differences |
Preset 1: Setup Contribution
Question: which labeled setups contributed to the scoped result, and how concentrated is each result?
Display: trade count, active days, net P&L, average and median outcome, gross wins and losses, costs covered, largest-trade share, and missing-label count. Profit factor can be shown when gross-loss coverage is valid, but no universal PF band turns a setup into an edge.
Do not infer: the highest in-sample setup will remain best. A tiny or outlier-dominated leader is a hypothesis. Use the filter-your-edge workflow to test an inclusion or exclusion rule without deleting the official baseline.
Preset 2: Session Performance
Question: how did outcomes and exposure differ across declared session buckets?
Display: bucket definition, timezone, trades, active days, net result, average/median outcome, cost coverage, and unclassified timestamps. The busiest session can own the most P&L without having the strongest typical trade.
Do not infer: a session label causes performance or that its internal hours are uniform. The hour-by-hour performance map adds resolution with timezone, uncertainty, and later-sample controls.
Preset 3: Weekday Breakdown
Question: is one weekday materially different after accounting for activity and concentration?
Display: trades, active dates, total and per-trade results, median, largest-day share, and strategy mix by weekday. A day with one event is not a recurring weekday effect.
Do not infer: “never trade Friday” or any other permanent rule from a calendar label. Inspect event timing, setup mix, and opportunity count. The weekday analysis guide supplies the deeper control checklist.
Preset 4: Post-Loss Timing
Question: what happened to entries opened soon after a recorded loss compared with other timed trades?
Display: the exact interval rule, eligible prior-loss count, matched entries, net and average outcomes, size change, quality labels, and missing timestamp coverage. TSB currently exposes an observed “within 15 minutes after a loss” group; that is a timing classifier, not a diagnosis of revenge motive.
Do not infer: emotion or causality from proximity alone. A planned strategy sequence, event cluster, scaling rule, or correlated instrument can create the same pattern. Review the actual trades before applying a cooldown.
Preset 5: High-Frequency Day Comparison
Question: how do days above the record’s pace threshold compare with other active days?
Display: threshold definition, active-day distribution, days above threshold, trades and P&L in each group, strategy mix, and sequence context. Current TSB server logic derives its overtrading threshold from the trader’s daily pace rather than a universal fixed cap.
Do not infer: “more trades caused the loss.” Volatility, opportunity volume, news, strategy choice, or losses earlier in the day may drive both activity and outcome. The overtrading analysis explains how to test a cap without rewriting history.
Preset 6: Long Versus Short
Question: did long and short entries show different distributions inside the same comparable scope?
Display: count, active dates, instruments, setup mix, net result, median, payoff distribution, and missing-direction rows. If one direction occurs mostly in a different market regime or symbol, the raw split is confounded.
Do not infer: a permanent directional identity. Test whether the gap survives within the same setup and instrument group, then check later trades before creating a long-only or short-only rule.
Preset 7: Instrument Contribution
Question: which instruments contributed, diluted, or concentrated the result after costs and exposure?
Display: trade count, active days, net and per-trade outcomes, cost coverage, setup mix, capital/risk context, and largest-trade share. Currency totals are not comparable when account currency or risk scale differs.
Do not infer: that the weakest historical symbol should be removed immediately. First test whether it reflects a discontinued setup, incomplete fee data, a different period, or an outlier.
Preset 8: Evidence Report Card
Question: is the record ready for a decision, and which evidence weakness matters most?
A composite score is useful as navigation, not truth. Keep its components visible: data coverage, outcome quality, risk/control evidence, label consistency, concentration, and stability. A grade without the component inputs creates false precision.
Display: component values, missing fields, calculation version, scope, and why the weakest component received its status. If a component is unavailable, show Not verified rather than silently averaging around it.
Do not infer: profitability, skill, or future outcomes from one grade. The report card should route the reader to the source trades and the next verification step.
The 10-Minute Sequence
When the data gate already passes, use this compact sequence. Stop the clock whenever reconciliation is needed; data repair is work, not diagnostic delay.
- Minutes 0–1: freeze scope and confirm coverage indicators.
- Minutes 1–3: read setup, session, and weekday contribution.
- Minutes 3–5: inspect post-loss timing and high-frequency days.
- Minutes 5–7: compare direction and instrument groups.
- Minutes 7–8: read the evidence report card and its missing inputs.
- Minutes 8–10: write one hypothesis, counter-evidence, reversible action, and review condition.
The time target controls analysis sprawl; it does not cap judgment. A material outlier, mapping gap, or strategy-version break should trigger a deeper review instead of a rushed action item.
Save the Presets as a Versioned Contract
A saved view is not reproducible if its filters, grouping logic, or missing-value behavior can change invisibly. Give each preset a short definition record containing its population filter, date basis, grouping key, outcome field, cost policy, unknown-value policy, and calculation version.
| Contract field | Example purpose | Failure it prevents |
|---|---|---|
| Population filter | Names the exact accounts and closed-trade states | Different presets silently analyzing different ledgers |
| Date basis | Entry, exit, or another declared timestamp | Weekday and session rows changing with the chosen clock |
| Grouping rule | Defines setup, session, pace, direction, or instrument buckets | Renamed labels rewriting the historical comparison |
| Outcome and cost policy | States gross/net P&L, fees, financing, and conversion | Comparing a cost-complete view with a partial one |
| Unknown policy | Keeps missing labels as a visible group | Improving a result by dropping inconvenient rows |
| Version | Links the output to the calculation used | A software or formula update being mistaken for performance change |
Export or screenshot the eight outputs with the decision note. The artifact should let a later reviewer answer four questions without reconstructing the dashboard: what data was eligible, what was missing, which difference was observed, and what action was authorized. Preserve the earlier artifact when the preset changes.
Read Conflicts Before Picking the Action
The presets will sometimes disagree. A session may be positive overall while its most common setup is negative; an instrument may look weak because it was traded mostly on high-frequency days; a post-loss group may look poor only in one strategy version. Conflict is not a defect—it is a reason to inspect the intersection before acting.
Prefer the explanation that uses the fewest unsupported assumptions and survives a comparison inside the same strategy, instrument, and period. If the intersection becomes too small, stop. Record both observations and choose no action until more evidence arrives. “No supported change” is a valid diagnostic outcome.
Hidden Deal-Breaker: Closing the Action Loop
Reading eight charts can feel productive even when nothing changes. The safeguard is not motivational language; it is a versioned decision record.
Choose one action because simultaneous changes obscure attribution and increase operational load. The action must name the eligible population, behavior, start, review condition, and rollback trigger. “Trade better on Fridays” is not testable. “For the next review window, shadow-record but do not take entries matching this exact Friday filter; revert if opportunity-cost or coverage evidence crosses the stated condition” is testable.
One action is a review constraint, not a claim that the largest historical leak is causal. If the later evidence does not support it, restore the prior policy and record the failed hypothesis. A clean rejection is useful progress.
How the Eight Presets Map to TSB
TSB is our product. The current product exposes the underlying evidence across Dashboard and Coach surfaces: contribution by setup label, performance by session and weekday, high-frequency day comparison, post-loss timing, Edge Score, scoped trade review, and analysis prompts. The guide’s eight-preset sequence is an editorial workflow over those surfaces; it is not a claim that one button completes every review in every account.
TSB recognizes 330 import profiles and has processed 600K+ imported trades. Those canonical figures describe ingestion scope and product scale. They do not describe your selected evidence set, prove an identified leak, or promise improvement.
Use the source-evidence links to inspect the rows behind an observation. Reconcile accounts, costs, timestamps, setup labels, and duplicates first. A lifetime-access route is available alongside the current plan rendered in the server-side provider card. If evidence is missing, the diagnostic should say Not verified—not fill the gap with an AI story.
Cadence, Sample Size, and Change Control
- Cadence: review after a meaningful batch of new evidence or a declared strategy/import change. Weekly, monthly, or quarterly can all be valid; choose before seeing the result.
- Sample size: no universal threshold guarantees a signal. Show count, active periods, dispersion, concentration, and uncertainty; scale claim strength to them.
- Versioning: preserve the baseline filter and calculation version. Never alter historical labels just to strengthen the chosen action.
- Multiple comparisons: eight presets generate many candidate findings. Treat the most dramatic one as a hypothesis until a later window supports it.
- Outcome boundary: a better process metric does not automatically mean better P&L, and historical P&L does not guarantee a repeat.
Our editorial evidence methodology defines how verified facts, first-party observations, inferences, hypotheses, opinions, and catalog gaps are separated.
Final Verdict: Eight Fast Questions, One Auditable Decision
The eight presets are powerful because they constrain the questions, not because they remove judgment. They make setup, timing, behavior proxies, direction, instruments, and evidence quality visible in a repeatable order.
The useful output is a short chain: valid scope → observation → alternative explanations → one reversible action → later comparison → keep, revise, or roll back. With that loop, ten focused minutes can beat an open-ended evening of chart hunting. Without it, eight attractive views are only diagnostic entertainment.
