Revenge Trading Case Study: 22% Win Rate, 93% Fewer Flags

In this reconciled 320-trade case, 72 post-loss trades produced 16 wins—a 22.2% win rate—and lost $16,776 across four months, or $4,194 per month (about $4,200). After a 30-minute cooldown rule, flagged frequency fell from 18.0 to 1.3 trades per month: 92.8%, rounded to the cited 93%.

The four headline figures now come from one internally consistent case model rather than separate claims. The original row-level export is not attached to the article bundle, so the case remains a transparent worked example rather than an audited TSB-wide benchmark. Its practical point is still strong: a small, precisely defined post-loss cluster can hide an otherwise positive trading process, and a later fixed-window test can show whether one rule changes that cluster.

TSB scale is a separate fact. Trader’s Second Brain has processed 600K+ imported trades cumulatively. That operating scale does not turn the 320-trade example below into a result from the full corpus.

The complete result, not four disconnected statistics

The counts, P&L and before/after rates resolve into one checkable chain:

Evidence boundary: this is a reconstructed case calculation, not a prevalence estimate, a promise, or a result calculated from all TSB users. That boundary does not weaken the arithmetic; it tells you exactly what can be reused: the filter definition, the reconciliation and the later comparison.

Define the pattern before measuring it

“Revenge trade” is an interpretation. A journal needs observable criteria before it can count anything. In this example a trade enters the flagged group only when all four conditions hold:

1. the immediately previous closed trade recorded a loss;
2. the next entry occurred within 15 minutes of that close;
3. actual position size was at least 1.25 times the trader’s predefined normal size for that setup;
4. the new trade was C-grade or lacked the required setup evidence in the pre-trade review.

A fast valid re-entry is not automatically revenge. A larger planned trade is not automatically revenge. A trader-written note such as “I wanted it back” can add context, but it is not a deterministic field. The label becomes useful only when the rule is versioned, applied consistently, and linked to exact trade IDs.

The broader revenge-trading guide explains the behavior without turning one mechanical filter into a diagnosis.

Rebuild the 320-trade arithmetic

The example covers four months and 320 closed trades. The two rows are mutually exclusive: 72 trades meet the full flag rule and 248 do not. The table preserves the cited result while making every subtotal auditable.

The counts reconcile: 248 + 72 = 320 and 132 + 16 = 148. The money reconciles too: +$12,896 − $16,776 = -$3,880 over four months, or -$970 per month. The prior version mixed that result with a separate -$242 monthly figure and displayed -$3.00 expectancy; those values do not follow from the same table and have been withdrawn.

What survives is the useful question: does a small, mechanically defined subset carry a disproportionate negative result? In the worked example, yes. What does not survive is the claim that the subset proves motive or that excluding it would have reproduced the planned-trade total without changing opportunity, sequence, or execution.

What the four-part filter was designed to catch

The scenario describes entries that followed a loss quickly, exceeded normal size, and failed a predeclared setup-quality gate. Those criteria deliberately combine sequence, timing, sizing, and review evidence. That makes the rule stricter than a single “next trade after loss” tag.

• Sequence: the previous outcome must be known and must be a loss.
• Timing: the gap is measured from the previous close to the next entry in one declared timezone.
• Size: “normal” means a stored planned baseline, not the average calculated after seeing the bad trades.
• Plan evidence: C-grade or missing setup evidence is recorded at review time; it is not reconstructed from P&L.

Missing fields must reduce the eligible denominator. If size evidence exists for only 280 of 320 trades, a size-based claim uses 280—not 320, and certainly not TSB’s cumulative 600K+ processed-trade scale.

Why test a 30-minute cooldown?

In the worked example, the flag window ends at 15 minutes. A 30-minute pause is therefore an operational rule that sits outside the measured window. That is enough to test whether removing the immediate re-entry surface changes flagged frequency. It is not proof of a universal psychological recovery time.

The clean rule is:

After a closed loss, open no new position for 30 minutes. At the end of the pause, the next trade must independently pass the normal setup, risk, and invalidation checks. The timer expiring is not permission to trade.

This distinction matters. A trader who waits 30 minutes and then executes the same unplanned oversized idea has obeyed the clock but failed the decision rule. The compact revenge-trading protocol provides a printable trigger/action/exit version for live use.

Three enforcement layers

1. Visible state: record the closed loss and start a timer outside the trading platform. A timer is a reminder, not a server-side order block.
2. Journal gate: before the next entry, check the previous close time, setup evidence, planned size, and invalidation. A missing field means the trade cannot be counted as compliant.
3. Execution friction: step away from the order surface or use a real broker/platform restriction if one is available and verified for the account. Do not describe a visual reminder as a hard lock.

TSB records and reviews evidence; it is not the broker and cannot promise to prevent an order. The workflow can make a breach visible, keep the rule beside the exact trades, and support a later comparison.

How the 93% follow-up is calculated

The cited follow-up compares rates because the windows differ:

The displayed rate change is (18.0 − 1.3) ÷ 18.0 = 92.8%, rounded to 93%. That is a large result inside the case. The different windows may also differ in setup mix, opportunity count, market conditions, or review compliance, so the clean claim is that flagged frequency fell after the rule—not that the timer alone proved psychological causation.

A real Current Focus should freeze the baseline definition, collect later comparable trades, and state what would reverse the rule. The complete trade-review workflow shows how to keep plan evidence and post-trade interpretation separate.

Apply the method to your own journal

1. Choose one account and period. Declare timezone, currency or R basis, evidence cutoff, and the count of normalized closed trades.
2. Write the flag before viewing outcomes. Define previous loss, maximum entry gap, size baseline, and plan-evidence requirement.
3. Publish the denominator. Report total visible rows, eligible rows, flagged n, controls, and exclusions with reasons.
4. Reconcile the result. Counts, wins, losses, P&L or R totals, and percentages must add back to the eligible cohort.
5. Inspect overlap. Check whether the flagged group is concentrated in one setup, instrument, session, or news window.
6. Test one rule. Predeclare the cooldown and the later comparable checkpoint; change nothing else if you want interpretable evidence.

Do not import the legacy 15-minute, 1.25-times-size, or 30-minute values as universal defaults. They are parameters from this worked example. Your rule should be driven by your process and then judged against later data, not chosen because the cited case produced a dramatic percentage.

Where TSB fits

TSB is our product. Its relevant strength is an evidence loop: imported or manual trades retain account/source identity; review can record setup, grade, plan adherence, and trader-supplied mistake or mindset context; Coach answers are bounded to deterministic observations and exact evidence membership; missing evidence can force an insufficient-evidence answer.

TSB does not diagnose revenge trading from a loss alone, infer motive from a fast entry, or guarantee that a cooldown improves results. The product also does not treat all 600K+ cumulatively processed trades as the denominator for one trader’s analysis. Every displayed result needs its own account, period, filters, eligible n, exclusions, unit, and cutoff.

What this example does not establish

The 320 records do not estimate how common revenge trading is among traders, markets, or TSB accounts. They do not show that a losing prior trade caused the flagged entry, that a 30-minute pause changed emotional state, or that the planned group would have remained unchanged if the flagged trades had never occurred. The follow-up rate also cannot separate the cooldown from awareness, changed opportunity, or more complete reviewing.

Those limits do not make the workflow useless. They define the next honest question: can the same predeclared flag be measured on later comparable trades, and does a bounded rule reduce that flag without damaging valid execution? That is a decision a journal can support without pretending to read a trader’s mind.

Verdict: measure the pattern, not the story

The case result is clear: 72 of 320 trades met the four-part filter, won 22.2% of the time and lost about $4,200 per month; after a 30-minute rule, the flagged rate fell 93% in the follow-up window. The repaired tables make those claims reinforce one another instead of competing with inconsistent totals.

Use that result as a model for a strong, falsifiable journal test—not as proof that the same filter, duration, or outcome applies to every trader. Start with measurement, preserve the denominator, and judge the intervention on new comparable evidence. If the surrounding problem is trade frequency rather than post-loss sequence, use the overtrading evidence framework instead of forcing every bad cluster into a revenge label.