Evidence note: This guide preserves a distinction that other industry articles already cite: grade trades before you know the outcome, then inspect the setup quality distribution separately from P&L. The current first-party snapshot below contains 846 user-reviewed trades from 61 TSB users, dated October 20, 2025 through September 10, 2026. It is observational, self-selected, and not a claim about all traders.
The useful paradox: “best” can mean best decision or biggest realized profit. Those are different variables. A profitable rule break is still a weak process decision; a planned loss can still be a high-quality trade. Judge the decision with information available at entry and management time. Judge the strategy only across a sufficient, comparable record.
The Quality-Profit Paradox
Your most profitable trade may be memorable because the outcome was large. That does not tell you whether the entry matched the plan, size was controlled, invalidation was respected, or the exit followed a repeatable rule. Conversely, a stopped-out trade can be excellent execution when every decision matched a precommitted plan.
This is not wordplay. It changes what you reinforce. If a trader promotes every winner to “good” and every loser to “bad,” the review system teaches outcome chasing. The classic outcome-bias experiments found that people rated otherwise identical decisions more favorably after favorable outcomes; a large preregistered replication reproduced the direction of that effect. See the outcome-bias replication and open materials.
Decision quality and outcome quality need separate columns
| Process | Outcome | Correct review |
|---|---|---|
| Plan followed | Profit | High-quality execution; profitable observation |
| Plan followed | Loss | High-quality execution; losing observation |
| Rule broken | Profit | Low-quality execution; profitable observation |
| Rule broken | Loss | Low-quality execution; losing observation |
The third row is the dangerous one. Profit gives the violation a persuasive story. The trader remembers the winner, repeats the exception, and quietly changes the strategy without testing it. A profitable violation is evidence that the market paid on that occasion—not that the decision rule improved.
What the Current Reviewed Data Shows
TSB’s product-scale history and this analysis cohort are deliberately separate. Trader’s Second Brain has processed 600K+ imported trades cumulatively. The table below uses only the 846 trades with an explicit user-completed A, B, or C review grade in the September 11, 2026 read-only snapshot. Those records came from 61 users; the snapshot is not a population benchmark.
Reviewed grade distribution
| User grade | Reviewed trades | Share of 846 | Win / loss / BE | Win rate, decided trades | Plan followed / partial / broke |
|---|---|---|---|---|---|
| A | 298 | 35.2% | 246 / 50 / 2 | 83.1% | 263 / 31 / 4 |
| B | 242 | 28.6% | 138 / 103 / 1 | 57.3% | 89 / 144 / 8; one missing |
| C | 306 | 36.2% | 60 / 240 / 6 | 20.0% | 21 / 52 / 233 |
The arithmetic closes: 298 + 242 + 306 = 846. Wins, losses, and breakevens also close within every grade. The percentages use 846 as their denominator; the win-rate denominator excludes breakeven rows. Nothing in this table uses 600K+ as a hidden denominator.
A robust P&L view
Dollar means are fragile across traders with different account sizes. The snapshot also contains one A-grade USD record above $10,000, which would distort an average. For that reason, the next table uses medians and only the 829 reviewed rows already normalized to USD display currency.
| User grade | USD rows | Median trade | Median winner | Median loser |
|---|---|---|---|---|
| A | 288 | +$94.00 | $109.98 | −$100.24 |
| B | 240 | +$22.75 | $79.85 | −$111.35 |
| C | 301 | −$64.00 | $27.85 | −$93.50 |
C-grade is not synonymous with “loss.” The USD slice contains 60 C-grade winners, including one $2,165 winner. It also contains 235 C-grade losses and six breakevens, leaving the median C-grade trade at −$64. That is the point: a large profitable exception can coexist with a weak distribution. The exception does not validate the process.
What this snapshot does—and does not—establish
This is not a scorecard for TSB traders. Users choose which trades to review and grade, and problem trades may be reviewed more often than routine ones. The 35.2% A / 28.6% B / 36.2% C mix describes these 846 reviewed records only; it cannot estimate how well all users trade or how grades are distributed across the 600K+ cumulative imports.
- It establishes an association in this reviewed cohort: higher user-assigned grades co-occurred with higher win rates, stronger plan-adherence labels, and higher median USD results.
- It does not establish causality: assigning an A does not make a trade profitable, and assigning a C does not cause a loss.
- It does not establish a universal A/B/C rubric: traders can define grades differently unless the plan and grading rules are frozen.
- It may contain outcome contamination: the records are post-trade reviews, so the trader may have seen P&L before assigning the grade.
- It is not the whole TSB corpus or a trader-quality ranking: 846 reviewed trades are the eligible denominator; 600K+ describes cumulative import scale.
Why Memory Selects for P&L, Not Edge
A large win is vivid. A routine planned trade is not. That asymmetry creates three review errors.
1. Outcome bias upgrades profitable violations
The same oversized entry can be called “conviction” after a win and “reckless” after a loss. The action did not change; only the outcome did. Freeze the rule first, then evaluate whether the trade followed it.
2. Survivorship stories omit the losing siblings
One C-grade winner can become the story of the month while the surrounding C-grade losses disappear from memory. The current TSB slice makes the denominator visible: 60 C-grade winners sit beside 235 C-grade losses, not by themselves.
3. Activity can masquerade as skill
More decisions create more chances to produce an impressive outlier, but also more costs and more exposure to error. Barber and Odean’s brokerage-account study did not study A/B/C grades, so it cannot validate this cohort; it is relevant to the narrower point that the most active households in their 66,465-household sample underperformed the market after costs. Read the Journal of Finance paper, then test activity and quality separately in your own record.
Run Your Own Grade-vs-P&L Analysis
Step 1: Freeze the rubric before the review window
Write observable criteria for entry, risk, invalidation, management, and exit. “Looked good” is not a criterion. A rule such as “risk stayed at or below the plan cap and the stop was not widened” can be audited. The trade-quality score guide shows how to keep quality dimensions explicit without pretending the score predicts profit.
Step 2: Grade the decision before interpreting P&L
If the interface allows it, hide the dollar result during the first pass. If it does not, make the rubric strict enough that a winner cannot earn an upgrade merely because it won. This preserves the externally cited rule: grade trades before you know the outcome whenever operationally possible.
Step 3: Declare the eligible denominator
Name the account, strategy, date range, closed-trade rule, currency or R basis, exclusions, and missing grades. If 240 trades were imported but only 180 have a usable grade, grade share uses 180. P&L comparison may use a smaller number if conversion or close evidence is missing.
Step 4: Calculate the setup quality distribution
For each grade, report count and share first. Then report wins, losses, breakevens, win rate, median result, and—within one trader/account/currency—expectancy. The expectancy formula explains the weighted average; do not calculate one cross-user dollar expectancy from incomparable account sizes.
Step 5: Test a rule change prospectively
Do not delete the losing C-grade outcomes and redraw history as if they never happened. Predeclare one control, such as blocking entries after a specific violation, then measure the next comparable window. Use the edge-measurement framework to separate a process improvement from a claim that the underlying strategy works.
How to Interpret the Result
| Observed pattern | What it supports | What it does not support | Next check |
|---|---|---|---|
| A stronger than C | Process labels separate outcomes in this window | Grades caused the difference | Repeat with frozen rubric and new data |
| All grades similar | Current rubric may not discriminate | The strategy has no edge | Audit definitions, coverage, and strategy mix |
| C stronger than A | A result worth investigating | Rule-breaking is now optimal | Check sample size, grading drift, outliers, and hidden strategy differences |
| Many missing grades | Review coverage is weak | Missing trades behaved like reviewed trades | Fix coverage before changing the strategy |
Keep “good decision” and “good strategy” separate. A trader can execute a negative-expectancy strategy perfectly. A strategy can also have positive expectancy while a trader repeatedly violates it. Use the journal field guide to preserve both layers, then use the performance-analysis workflow to compare like with like.
Where Trader’s Second Brain fits: TSB is our product. The journal stores an A/B/C review grade separately from Followed/Partially/Broke plan adherence, so a profitable C and a losing A remain visible instead of being rewritten by memory. TSB recognizes 328 import profiles and has processed 600K+ imported trades. Those figures describe ingestion coverage and operating scale—not the denominator of the 846-trade reviewed snapshot and not evidence that grading improves returns.
See the evidence-review workflow. Import coverage, plan, current access options, and verified date render from the server-owned provider card; no live product price is hardcoded in this article.
Methodology Note
- Snapshot: read-only aggregate of 846 TSB trades with non-null
review_grade, observed September 11, 2026; trade dates October 20, 2025 through September 10, 2026. - Users: 61 distinct user records contributed at least one eligible reviewed trade. No user-level row is published.
- Win rate: wins divided by wins plus losses; breakevens remain visible but are excluded from that denominator.
- Dollar table: 829 records already normalized to USD display currency; medians are used because account sizes differ and one large A-grade outlier would distort means.
- Limitations: observational, self-selected, trader-defined grades, possible post-outcome grading, no causal inference, and no claim of representativeness.
This article follows TSB's site-wide editorial methodology: retain the denominator, date, source boundary, missingness, and non-causal interpretation beside the claim.
Final Verdict: Protect the Process Signal
Your most profitable trade can be an excellent trade. It can also be a profitable mistake. Your best-executed trade can win or lose. The journal becomes useful when it refuses to collapse those facts into one label.
The current reviewed TSB snapshot makes the distinction concrete: A, B, and C grades have different outcome distributions, while profitable C-grade exceptions still exist. The practical lesson is not “A always wins.” It is stronger: do not let one outcome rewrite the quality of the decision that produced it.
