There is a dangerous comfort in a high win rate. It feels like proof that the method works, until the math shows that the average loss is much larger than the average win. This FTMO Challenge account is a useful case study because the evidence is clear: 743 imported MT4 execution records, 100% concentration in XAUUSD, and a final net P/L of -$885.33 across the closed-record sequence.
The account was operating under FTMO Challenge constraints: a 10% profit target, -5% daily max drawdown, and -10% overall max drawdown. It did not fail because trades were rare or because the operator had no winners. It failed because the realized pattern of returns did not respect the account’s loss boundary. The central lesson is not about one bad trade. It is about process design, position sizing, and the cost of letting recovery behavior control the sequence.
Observation: the account had a positive hit rate but negative economics
The raw statistics are straightforward. Win rate was 60.57%, gross profit was $747.70, gross loss was -$1,633.03, and profit factor was 0.458. Average win was $1.66 while average loss was -$5.57. Expectancy per trade was -$1.19. In other words, the account could be right more often than wrong and still lose money because the losses were structurally larger than the gains.
This is a classic high-win-rate, negative-expectancy profile. The approved interpretation matters here: a high win rate with negative expectancy can be consistent with taking profits quickly while allowing some losses to become much larger. That does not prove intent, but it does describe the outcome. The account also showed a worst 10% loss share of 64.11% and a worst 1% loss share of 29.34%, which means the tail mattered materially.
Realized cumulative P/L from closed records. This shows the closed-trade outcome only, not an equity curve.
Closed-record drawdown reconstructed from realized outcomes. This is a realized-loss view, not an intraday mark-to-market series.
Explanation: concentration and overlap amplified the same market exposure
All 743 records were XAUUSD. That concentration may reflect a specialist mandate, but it also amplified exposure to one market regime. When one instrument dominates the entire sample, the strategy’s true behavior becomes tightly linked to that market’s volatility, spread conditions, and session timing.
The trade log also showed 99 same-direction overlapping entries. That pattern may represent DCA or a planned multi-entry execution; the data alone cannot distinguish them. What it does tell us is that exposure was being layered instead of remaining isolated. The result was predictable in one sense: the account was vulnerable to adverse streaks, especially when losses were not cut decisively.
The size analysis reinforces this point. The smallest tier had 362 trades and a high win rate of 80.11%, but still lost $121.62. The Tier 2 bucket produced -$436.23 across 368 trades. The larger buckets were too sparse to rescue the overall result. This is what negative expectancy looks like in practice: more wins do not automatically compensate for loss asymmetry.
Monthly realized P/L from closed records. It highlights how the sequence deteriorated across the sampled months.
Recorded position size through time. The chart helps compare sizing behavior with realized outcomes.
Implication: higher activity made the account worse, not better
The account did not improve through more trading. Four high-activity days, defined as at least 47 trades, averaged -$2.00 per trade versus -$0.58 on normal days. The higher-activity bucket also accounted for 627 trades and -$842.58 in net loss. This is an overtrading signature: unusually high activity coincided with worse outcomes.
Frequency did not create edge. It created friction. Daily pnl volatility was $107.26, median trades per active day was 19, and there were 28 active days in total. Rapid re-entry was also common: 255 rapid post-loss reentries and 366 rapid post-win reentries. The rapid post-loss reentry rate was 87.03%, and the rapid post-win reentry rate was 81.33%. That is not automatically a flaw, but it becomes one when the follow-up trade inherits the prior trade’s emotional and risk burden instead of a clean setup.
The expectancy after a win was -$0.99, and after a loss it was -$1.48. The next trade was worse after losses than after wins. Approved interpretation is careful here: performance weakened after losses; recovery-seeking, changing conditions, and strategy sequencing are competing explanations. The data do not prove why, but they do show that the sequence mattered.
Daily trade activity overlaid on realized P/L. It shows how activity clustered around the losing sequence.
Rolling trade expectancy. This helps reveal whether the edge, if any, was stable or fading.
Explanation: time-of-day and weekday behavior were not uniform
Some hours were meaningfully better than others. Hours 8, 9, 10, 12, 19, and 20 showed positive expectancy, while hour 18 was particularly poor at -$18.00 expectancy across 23 trades. Hour 13 also stood out negatively at -$6.60 expectancy. The pattern suggests that timing mattered more than average trade count might imply.
Weekday results were similarly uneven. Wednesday and Thursday were positive in net terms, while Monday was sharply negative at -$768.32. Again, this does not prove causation, but it does tell the operator where to look. If a system is profitable only in certain hours or on certain days, it is not a generic edge; it is a conditional edge that needs strict boundaries.
The holding time data adds another layer. Median winner hours were 0.03 and median loser hours were 0.01. Most positions were very short-lived, with a holding hours median of 0.02 and a p90 of 0.46. That means the strategy was operating in a fast, noisy part of the market where costs and spread matter, especially when execution is concentrated in XAUUSD.
Distribution of closed-record outcomes. This makes the asymmetry between small wins and larger losses easier to see.
Rapid re-entry rates after losses versus wins. This shows how frequently the next decision arrived before the prior one was fully digested.
Implication: the lesson is risk design, not regret
The operator’s reflection is consistent with the evidence. The account failed by breaching max daily drawdown because the bot was not programmed to prevent such an event. The operator also stated that DCA was abandoned because it could not actively control risk size relative to target profit, and because the pursuit of income was penalized by large drawdowns. That is a practical conclusion, not a moral one.
For sophisticated traders and investors, the lesson is simple but unforgiving. A strategy can have a respectable win rate, even a streak of profitable days, and still be unfit for a challenge account if the tail loss is uncontrolled. The real question is not whether the method can generate a win. It is whether the method can survive the bad sequence that eventually arrives.
A useful risk framework from this case would be: limit same-direction stacking, predefine maximum exposure per instrument, impose a hard guardrail after a loss sequence, monitor expectancy by session, and separate signal quality from trade frequency. On a challenge account, survival is the first objective. Profit comes only after that. Without a designed loss boundary, the account can look active, look engaged, and still fail in the only way that matters.
The broader investing principle is familiar. Compounding depends less on being right often than on not being wrong too much at the wrong time. This account is evidence that the market can extract money from a system that confuses activity with control. The fix is not to trade less for its own sake. The fix is to make every additional trade earn its right to exist within a known risk budget.
Readers who want more account-level case studies on expectancy, drawdown control, and execution quality should review related analyses before scaling any strategy into a live or funded environment.