Across 1,000 Monte Carlo trials, A-3 cleared the FTMO Phase-1 floor with a 95% CI of 71–84%. Drawdown stacking simulation held under 8.2% max DD against the 10% rule. Sharpe-ratio lower bound of 1.42 across 3,517 simulated trades over a 3-year backtest period.
Recommended for traders prioritizing consistency over peak return. Primary firm match: FTMO 2-step Phase 1 on highest CI lower bound. Conservative on every margin — same-seed reproducible, audited per the Pass Lab methodology.
Why this portfolio?
Portfolio A-3 is the recommendation because it had the highest CI lower bound across the Pareto-optimal candidates, not the highest point estimate. A point estimate of 91% with a wide interval of [54%, 91%] would look impressive on a chart and fail the same trader on the same backtests in a different intraday trade order. A-3's tighter [71%, 84%] is more stable across walk-forward windows — and tighter is what survives live execution where you can't re-pick the seed.
The drawdown signature is what makes A-3 different from neighbouring Pareto candidates. A-1 had a higher Sharpe point estimate (1.78 vs 1.42) but wide swings — its 5th-percentile path crossed the FTMO 5% daily-DD ceiling on three of 1,000 Monte Carlo trials. A-3 has shallower drawdowns, spread across uncorrelated EAs that don't lose together. The compounding of small, uncorrelated drawdowns is what a "shared-balance simulator" actually catches — and what merged-trade-list tools miss entirely.
The brand promise is the lower bound, not the best case: this is the worst run we found, not the best. If the CI lower bound clears the firm's pass threshold, the portfolio is the primary match — even if a different portfolio's point estimate looks higher on the marketing graph. The full methodology is documented at Pass Lab Methodology.
The composition
Portfolio A-3 combines five EAs across five major and minor pairs. The selection prioritises low cross-EA correlation (so drawdowns don't stack on the same day) and roughly equal solo Sharpe so no single EA dominates the portfolio's risk profile.
| EA | Pair | Lot type | Weight | Solo Sharpe | Solo Max DD |
|---|---|---|---|---|---|
| Trend-Bot v3 | EURUSD | Dynamic | 28% | 1.31 | 6.4% |
| Range-Snap RX | GBPJPY | Fixed | 22% | 1.18 | 7.2% |
| Carry-Edge 2 | AUDUSD | Dynamic | 19% | 1.04 | 5.9% |
| Breakout-Hour | USDJPY | Fixed | 17% | 1.27 | 7.8% |
| Mean-Revert FX | EURGBP | Dynamic | 14% | 0.96 | 5.1% |
Mean cross-EA correlation is 0.18 — weak positive — which is the headline win here: when one EA is in drawdown, the others have a 4-out-of-5 chance of being flat or up. The portfolio's combined max DD (8.2%) is below every individual EA's solo max DD because losses don't co-occur. That's the drawdown-stacking math working in your favour, not against you.
Mean cross-EA correlation is the most under-counted metric in EA portfolio sales pitches. A portfolio of "5 trend-following EAs across 5 majors" sounds diversified — until 2024-08-05 happens and every trend system on every major loses on the same Monday. A-3 deliberately mixes trend (Trend-Bot, Breakout-Hour), range (Range-Snap, Mean-Revert), and carry (Carry-Edge) to avoid that single-day-stacking failure mode.
Robustness
Robustness in this report is three numbers: the per-trial pass-rate distribution from 1,000 Monte Carlo replays, the walk-forward confidence interval across 24 sliding windows, and the Sharpe ratio bootstrap CI from 5,000 resamples. Each captures a different kind of "what if the trade order had been slightly different".
Monte Carlo: 1,000 trials
The same 3,517 trades were replayed 1,000 times with within-day order shuffled. Within-day shuffling preserves daily clustering (so daily DD remains realistic) but captures path-dependence: the same trades in different intraday order can pass or fail the daily-DD rule differently.
Distribution of pass-rates across 1,000 within-day-shuffled replays. The two dashed lines are the bootstrap 95% CI bounds. Mode falls at ~78% with both tails inside the firm's drawdown ceiling.
Walk-forward: 24 windows
A 30-day rolling window with 15-day stride generates 24 walk-forward observations across the 3-year backtest. Each window's pass-rate is computed independently; the dispersion across windows is what the bootstrap captures.
24 overlapping 30-day walk-forward windows (15-day stride). Every window is independently inside the bootstrap CI band, no obvious time-of-year clustering or regime-shift collapse.
Sharpe ratio: bootstrap CI
Sharpe ratio across 5,000 bootstrap resamples of the trade list: point estimate 1.61, 95% CI [1.42, 1.78]. The lower bound (1.42) is what the report headlines — same principle as the pass-rate: anchored on the conservative side, not the marketing side. Sharpe values above 1.0 are normally considered tradeable; A-3's lower bound clears that floor with margin.
Risk profile
The single most-misleading statistic in shared-account EA portfolios is "max drawdown" computed from a merged trade list. When two EAs lose the same week, the merged-trade-list approach reports the sum of their individual drawdowns reduced by any overlap — which under-states the real shared-account drawdown by 15-30%. FXOptimize replays trades chronologically against a single shared balance, so every dollar lost by one EA is a dollar of equity unavailable to another.
Drawdown stacking is real. A merged-trade-list shows A-3's max DD as 6.7%. The shared-balance simulator shows 8.2%. The difference (1.5 percentage points) is the drawdown-stacking premium — exactly what would have hit your account on a real funded challenge during the worst 30-day window.
Max-drawdown distribution across the same 1,000 trials. p95 lands at 8.2%; the long tail decays well before reaching the 10% FTMO ceiling. No trial breached the 5% daily-DD rule on any single day.
The headline number 8.2% is the 95th percentile of trial max DDs — i.e. the worst run we found in 95% of trials, conservative. It is not the average and not the best case. The remaining 5% of trials are in the long right tail (8.2% to ~10.4%), and even those did not breach the 10% rule by enough to reach the failure threshold across full evaluation windows.
Pass Lab cross-firm
Pass-rate against the V1 firm catalog. Each row is the same Portfolio A-3 evaluated against that firm's exact rule profile (drawdown type, profit target, daily DD limit, minimum trading days, consistency rules where applicable). Ranked by CI lower bound, not point estimate.
| Firm / Challenge | CI Low | Point | CI High | Match |
|---|---|---|---|---|
| FTMO 2-step Phase 1 | 71% | 77.5% | 84% | Strong · primary |
| FundedNext Stellar 2-step | 62% | 70% | 78% | Moderate |
| FundingPips 2-step | 58% | 64.5% | 71% | Moderate |
| The Funded Trader Standard | 54% | 61% | 68% | Moderate |
| FXIFY 2-step | 49% | 56% | 63% | Weak |
| The5ers Hyper Growth | 42% | 50% | 58% | Weak |
| FTMO 1-step (50% Best Day) | 38% | 45% | 52% | Weak |
| Goat Funded Trader Standard | 31% | 39% | 47% | Weak |
Primary match: FTMO 2-step Phase 1. Highest CI lower bound at 71%. The single firm where A-3 clears the Strong threshold (≥70% on the lower bound). FTMO 1-step lands lower because the 50% Best Day rule fails portfolios with one dominant outlier day — Trend-Bot v3 had four such days across the 3-year backtest, and the rule fires deterministically on each.
The ranking by CI lower bound is the recommendation logic. A different portfolio (B-2, not shown here) had a higher FundedNext point estimate (76%) than A-3's (70%) — but B-2's CI was [54%, 76%], visibly wider. A-3 is the more stable backtest match for FundedNext on the lower bound, even though B-2's point estimate looks better in isolation.
What's not shown
For the same audit-grade reason the page surfaces conservative numbers, here's the honest list of what this sample doesn't claim:
- This is a sample, not a recommendation for trading. Portfolio A-3, the EAs, the pairs, and the numbers are illustrative — synthesised to demonstrate what an FXOptimize report looks like. Real reports use your backtest data, not this one.
- Backtest is not live execution. Real trading adds slippage, broker latency, symbol restrictions, and news-window suspensions that the simulator does not fully model. The CI captures the variance from the trade-shuffling Monte Carlo; it does not capture broker-execution variance, which can be material on lower-spread pairs and during news windows.
- Past performance is not indicative of future results. The 71-84% CI is a backtest pass rate, not a prediction of your real funded-account success rate. Markets change; every model has a lifecycle. Pass Lab is a research/analytics tool, not financial advice.
- Synthetic data caveat. The trade-by-trade history that produced these numbers is synthetic for illustration. Numbers that anchor the report (71-84%, 8.2%, 1.42) are designed to be realistic but are not the output of any real EA trading any real account.
What you should NOT do: read this page, assume A-3 is a real portfolio you can buy or copy, or use these numbers in any external claim. The portfolio name is a placeholder. The EA names are placeholders. The numbers are designed to look real, not be real. Run FXOptimize on your own backtests for real numbers.
Generate your own research note
Upload your MT4 / MT5 backtest reports. FXOptimize parses them in your browser, runs the same Pareto + Monte Carlo + bootstrap pipeline, and surfaces the same kind of report — with your actual numbers, not synthetic ones.
Upload your backtests →Free tier — 3 EAs, 1 analysis per month. No credit card.