7 Prop Firm Evaluation Mistakes That Blow Accounts
Most failed evaluations aren't a strategy problem — they're a deployment problem. If the same general approach has failed more than once, one of these seven mistakes is almost certainly why.
A prop firm evaluation rarely fails because the underlying strategy has no edge. It fails because the strategy gets deployed in a way that ignores the evaluation's specific mechanical rules — trailing drawdown, contract size, news exposure. Fix the deployment and the same strategy often passes on the next attempt.
Mistake #1: Backtesting without the evaluation's actual constraints
A strategy that shows a strong annual return in a plain TradingView backtest can still fail an evaluation in its first week, because the default backtest never modeled a trailing drawdown, a session restriction, or a minimum trading-day count. It was optimized for a different scoreboard than the one the evaluation actually uses. Our guide to passing the Apex evaluation walks through adding those constraints to a backtest before an eval fee is ever spent.
The fix — build the evaluation's rules into every backtest run:
- Track a rolling equity peak and flag any bar where the drawdown from that peak exceeds the eval's limit
- Apply the actual session filter to historical data, not just live trading
- Count distinct trading days the strategy fired on, and confirm it clears the firm's minimum
- If the account has a consistency rule, simulate it — track the best day as a share of total profit
Mistake #2: Trading the full-size contract during the evaluation
Every major index and commodity future has a micro counterpart — MES instead of ES, MNQ instead of NQ, MCL instead of CL, MGC instead of GC — sized at roughly a tenth of the full contract's dollar risk per point. Running the full-size contract on an evaluation multiplies exposure per tick with no offsetting benefit during a phase that's purely about proving compliance, not maximizing size.
On a 50k Apex evaluation with a $2,500 trailing drawdown: a 25-point adverse move costs about $50 on 1 MNQ but roughly $500 on 1 NQ — $20 × price versus $2 × price. That's a meaningful chunk of the trail on a single trade with the full-size contract, for identical price action. There's essentially no scenario where trading full-size makes sense before an account is funded.
Mistake #3: No daily loss kill switch
Automated strategies chain losses like any other system — a strategy with a 55% win rate will statistically produce runs of five to seven losers in a row. That isn't a malfunction, it's ordinary variance. Without a daily cap, that run plays out in full and can take the trailing drawdown with it.
The fix — build a kill switch into every script:
- Track a running daily P&L variable from the session's opening equity
- Once that variable reaches roughly 40% of the account's trailing drawdown limit, halt new entries for the rest of the session
- Reset the counter at the next session's open
Daily kill switch — Pine Script snippet
Drop this into an existing strategy and gate every entry condition with not tradingHalted.
// ── Daily loss kill switch ─────────────────────────────────────────
// Halts new entries once the session's running loss hits the limit.
// Resets automatically at the next session open.
dailyLossLimit = input.float(1000.0, "Daily Loss Limit ($)")
isNewSession = ta.change(time("D")) != 0
var float dayOpenEq = na
dayOpenEq := isNewSession ? strategy.equity : dayOpenEq
dailyPnl = strategy.equity - nz(dayOpenEq, strategy.equity)
tradingHalted = dailyPnl <= -dailyLossLimit
// usage: add "and not tradingHalted" to every entry condition
// if longCondition and not tradingHalted
// strategy.entry("L", strategy.long)
plot(dailyPnl, "Daily P&L", color = dailyPnl >= 0 ? color.green : color.red)
hline(-dailyLossLimit, "Kill Trigger", color = color.red, linestyle = hline.style_dashed)Mistake #4: Trading through scheduled news events
CPI, NFP, FOMC, PPI, and jobless claims releases can move ES and NQ dozens of points within seconds. An indicator-based strategy has no real edge inside that window — it's pure noise risk, and on a trailing-drawdown account a single news-driven candle can be enough to breach the threshold outright.
Common high-impact release windows for US futures traders:
- 8:30 AM ET — CPI, NFP, PPI, jobless claims
- 10:00 AM ET — ISM, consumer confidence
- 2:00 PM ET — FOMC rate decisions (eight times a year)
- 2:30 PM ET — FOMC press conference
The fix — code a hard blackout window that blocks new entries for a few minutes before and roughly fifteen minutes after any scheduled high-impact release, driven by an economic calendar reference rather than a guess at the timing.
Mistake #5: A strategy tuned for the wrong market regime
A trend-following system performs well when the market trends. A mean-reversion system performs well when it ranges. An evaluation runs in real time against whatever regime happens to show up — and markets alternate between the two unpredictably.
A strategy backtested only against a strongly trending stretch, then deployed into an evaluation that opens during a choppy period, will underperform until the regime matches again — and by then the drawdown limit may already be gone.
The fix — test across at least one clearly trending period, one choppy/ranging period, and one period of elevated macro volatility. A strategy that only clears one of the three isn't ready to sit on an evaluation account yet.
Mistake #6: Ignoring slippage in the backtest
TradingView's backtester defaults to zero slippage — every fill happens exactly at the signal price. Live futures trading isn't like that. MES and MNQ typically see a tick or two of slippage per fill; CL and GC can see more on fast bars.
For a strategy trading a hundred times a month, even a couple of ticks of slippage per trade adds up to a real monthly cost — one that can erase a large share of a strategy's edge if the average trade profit is small to begin with. A backtest that looks profitable with zero slippage can be marginal or worse once realistic fills are modeled.
The fix — set slippage to at least two or three ticks per trade in the strategy's properties before trusting any backtest result enough to run it on a paid evaluation.
Mistake #7: Scaling contract count before the strategy has proven itself live
The most common reason funded accounts (not just evaluations) get blown is scaling up size before the strategy has shown it holds up outside a backtest. A backtest demonstrates hypothetical edge. Ten to twenty live trades demonstrate real-world edge. Fifty or more start to carry statistical weight.
The temptation is to run five or ten contracts the moment an account is funded, because the account size technically supports it. That skips the part where live performance is confirmed to actually resemble the backtest — which it never does perfectly.
The fix — start at the smallest contract count that makes the strategy meaningful, then add one or two contracts after every ten profitable live trades rather than jumping straight to a "full" size.