If you’ve searched for an EA to pass a prop firm challenge, you’ve already found ten lists that all say the same three things: high win rate, low drawdown, verified results. What none of them show you is how “low drawdown” was measured, over what data, or whether the number that got printed on the sales page is the worst case or just the one that happened to show up in the backtest.
We build EAs differently: an AI research engine generates thousands of strategy variations, runs every one through the same automated statistical gauntlet, and only the tiny fraction that survive ever reach a real account. Nothing below is a strategy pitch — we don’t publish entry logic, because entry logic is not the moat. The moat is that almost nothing survives the process, and we can show you the process.
At a Glance
| What matters for a challenge | PropGauntlet EA |
|---|---|
| Live/verified evidence | None publicly yet — new product. Stated plainly, not buried. |
| Pass-rate visibility | Simulated per-attempt pass odds — ~63% (Steady, 1.5% risk) / ~52% (Balanced, 2.0%) / ~40% (Turbo, 2.0%) — Monte Carlo, dev data, explicitly not a guarantee. |
| Disclosed max drawdown | Soft stops built in at 3.5% daily / 8% total — below FTMO’s own 5%/10% limits by design — plus the Monte Carlo 95th-percentile tail drawdown published below. |
| Disclosed methodology | Full 12-step battery published, including what fails at each step. |
| Free to verify yourself | Free Strategy Tester build and free unlimited demo forward testing (FxPro, AMarkets or xChief) — no license key, no signup for either. |
| What actually trades | Three index legs — SP500 (core), Nasdaq, and DAX (secondary). Only SP500 cleared the full robustness battery — the preset selector sets how much weight the other two carry (halved on Steady, full on Turbo), and Funded mode drops them entirely. |
| Price | $297 / year, 5 account activations. |
- Full 12-step testing methodology published, not just results
- Free backtest (Strategy Tester) and free unlimited demo forward-testing
- Soft stops set below the firm’s actual hard limits
- No grid, no martingale, no hedging
- No public live track record yet — new product
- Crypto is the only payment method
Can an EA Actually Pass a Prop Firm Challenge?
Honestly: sometimes, never certainly. Anyone promising a guaranteed pass is selling a story, not a strategy. What an EA can do, if it’s built and tested properly, is give you a repeatable, unemotional process with a measurable probability of passing — and let you decide, in advance, whether that probability is worth the entry fee.
That’s the number we publish: per-attempt pass probability, from thousands of simulated challenge attempts, at a stated risk-per-trade. Not a promise. An odds sheet. If a vendor won’t show you that number, or shows you total return instead, that’s the tell.
Why Most “Low Drawdown” EAs Still Fail Prop Firm Challenges
A few years ago we reviewed a stack of popular EAs for their drawdown numbers — that write-up is still live — and the honest conclusion was: some of them really do have low headline drawdown, and still aren’t built to pass a challenge. Two reasons keep showing up:
- Headline drawdown isn’t the number that kills a challenge. FTMO-style accounts fail on a daily loss limit (typically 5%) far more often than on the total loss limit (typically 10%). A strategy can have a gorgeous long-run equity curve and still die because one bad day, or two trades that both close the same session, blew the daily cap. Almost no vendor backtest even models a daily limit — they model the account curve, not the calendar.
- Average drawdown isn’t worst-case drawdown. The historical max drawdown in a five-year backtest is one sample from a distribution. The real question is the tail: if you ran that same edge over a thousand plausible alternate histories, what’s the 95th-percentile worst case? That’s a Monte Carlo question, not a “look at the chart” question, and it’s the one most sales pages skip.
Our approach treats the daily limit as the primary constraint, not an afterthought, and reports tail-risk, not just the historical curve. Below is exactly how.
What We Actually Check Before an EA Touches a Live Challenge
Every candidate strategy that reaches this stage has already survived an AI-driven screening process that discarded far more ideas than it kept (more on that below). What’s left goes through the same fixed battery of statistical tests — the same 12 checks, every time, no exceptions, no picking the test that flatters the result:
-
1
Data hygiene
A minimum of real historical price history — 5 to 8 years depending on timeframe, more for slower swing-style setups that generate fewer trades per year. No backtest starts on a window too short to mean anything.
-
2
In-sample screen
The strategy must show positive expectancy and a minimum profit factor on its training data before any more compute is spent on it. This kills the overwhelming majority of AI-generated candidates immediately, cheaply.
-
3
Out-of-sample validation
A second, independent slice of data the strategy never saw during screening. Still not the final exam — just the first checkpoint that has to be cleared before more time is invested.
-
4
Real-cost stress test
Re-run at 1.5× and 2× real spread and slippage — measured from live broker tick data, not assumed — and it has to stay profitable. One broker price change shouldn’t be enough to erase the edge.
-
5
Walk-forward validation
The data is split into six rolling folds. At least four of the six have to be independently profitable. One lucky stretch of history isn’t enough to pass.
-
6
Year-by-year consistency
At least 70% of calendar years must be profitable, and no single year may contribute more than 40% of total profit — the same instinct as FTMO’s own “best day” rule, applied yearly.
-
7
Regime check
A strategy that claims to trade trends has to actually be profitable in trending conditions specifically — not just profitable overall while secretly losing in the regime it claims to exploit.
-
8
Plateau / overfitting check
Every numeric parameter is nudged ±10–20%, and the edge has to survive on a plateau of nearby settings, not evaporate at one input’s rounding error. We report the plateau’s center, never the historical best single setting.
-
9
Monte Carlo tail-risk simulation
Thousands of resampled trade sequences from the strategy’s own history, to estimate the 95th-percentile worst-case drawdown and probability of ruin — the range of drawdowns that plausibly could have occurred.
-
10
Locked hold-out + statistical correction
A final slice of data — most recently 2025 through mid-2026 — is locked away and read exactly once. A Deflated Sharpe Ratio raises the bar based on how many candidates were tried globally, correcting for chance results.
-
11
FTMO-rule challenge simulation
Thousands of simulated challenge attempts, block-bootstrapped from real daily P&L, checking daily/total loss breach, profit target, minimum trading days, and the best-day rule — swept across risk levels.
-
12
Single-trade worst-case guard
The single worst adverse move any one trade has ever made is checked against the daily loss limit at every risk level. If it trips, that risk level is flagged and excluded from what we recommend.
No step in this list is optional, and no strategy skips a step because it “looks good.” That’s the entire difference between an odds sheet and a sales pitch.
The AI Engine Behind It
We didn’t hand-pick this strategy from a hunch. Our research engine generates and mechanically stress-tests thousands of strategy ideas and parameter variations — different entry filters, different sessions, different instruments, different exit logic — and runs every single one through the exact 12-step gauntlet above. The overwhelming majority die at step 1 or 2, cheaply, before any real compute is spent on them. Only a tiny handful ever reach the locked hold-out data, and most of those still get killed there. What survives to reach a live account is the rare exception, not the rule — and we can show you the graveyard, not just the survivor.
Free Backtest — Try Before You Trust It
You don’t have to take any of this on faith. The Strategy Tester build is free for everyone, no license key, no signup — load it in MetaTrader’s own Strategy Tester and run the exact logic yourself against your own broker’s data. Demo forward testing is free and unlimited too, on a demo at FxPro, AMarkets or xChief. The only thing gated behind a paid license is running it on a live or prop-firm account.
Three Ways to Run It: Steady, Balanced, Turbo
The EA ships with three presets, each trading a different balance of speed against pass probability — set once, no discretionary tweaking required. Numbers below are per-attempt Monte Carlo odds from our own development data (FTMO 2-step, 2019–2025), with the core leg additionally checked against a locked hold-out running through mid-2026 — not a promise of future results:
| Preset | Risk / Trade | Simulated Pass Odds (per attempt) | Median Time to Funded |
|---|---|---|---|
| Steady | 1.5% | ~63% | ~5 months |
| Balanced (default) | 2.0% | ~52% | ~2.5 months |
| Turbo | 2.0%, full weight | ~40% | ~5 weeks |
Read that table the way we’d want you to: Turbo is faster because it fails more often, not despite it. Steady is slower and boring on purpose. That trade-off is the whole product decision, and it’s yours to make, not ours to hide.
Steady, Balanced, or Turbo — pick the odds you want to trade
Per-attempt pass probability from Monte Carlo simulation on 2019–2025 development data, core leg hold-out-checked through mid-2026. Not a promise — an odds sheet.
Every line is one simulated phase-1 attempt real Monte Carlo output, dev data
Each line is a genuine attempt from the same Monte Carlo that produced the odds above — block-bootstrapped from the strategy’s real trades on 2019–2025 development data, run under FTMO’s phase-1 rules (+10% target, 5% daily loss, 10% total). Nothing is drawn by hand and nothing is smoothed. The sample shown fails at the same rate the full run did. Dashed red line: this preset’s daily soft-stop, set below the firm’s actual limit. Why the chart’s pass rate beats the stat row: the chart is phase 1 only; the headline figure is the whole challenge, and you have to clear phase 2 after this to get funded.
Why the two secondary legs (strategies) are on by default
This EA doesn’t run one strategy — it runs three, one per index (SP500, Nasdaq, DAX), and each leg has a different level of proof behind it. What actually moves between the three preset rows above is how much weight the two secondary legs carry, and they’re on by default in every preset, including Steady.
The SP500 leg is the one we trust most: it’s the only leg that passed our complete verification process, including a check against data that was locked away and never used while building it. The Nasdaq and DAX legs are profitable in our own testing too, but we were not able to verify them with the same rigor, and they have less history behind them than the SP500 leg, which we’ve tested back to 2019.
We shipped them on anyway, deliberately. They are what makes a challenge clearable in weeks rather than half a year, and hiding them behind a switch most buyers would never find would just be a quieter way of making the same decision. So the presets expose the dial instead:
| Preset | Core leg (SP500) | Secondary legs (Nasdaq, DAX) | What you’re choosing |
|---|---|---|---|
| Steady | 0.8 weight | 0.5 each — halved | Slowest, leans hardest on the verified leg |
| Balanced | 1.0 | 0.6 each | The default compromise |
| Turbo | 1.0 | 1.0 each — full | Fastest, most exposed to the unverified legs |
If you want the verified core only, that is exactly what Funded mode is: flip the Mode input and the EA drops both secondary legs and cuts risk to 0.75%, leaving only the hold-out-verified core. You flip it yourself — the EA does not detect that you’ve been funded — and you should, because the strategy that gets you funded fastest and the one that keeps you funded longest are not the same strategy.
The widget above is the honest version of this argument: switch the tabs and the pass odds move with the weights. That is the trade-off, priced.
What We Won’t Pretend
- Only the core strategy is verified against locked, never-touched out-of-sample data.
- The secondary legs are odds-positive but not independently battery-verified — and they ship enabled in every challenge preset, Steady included. The preset changes how much weight they carry, not whether they trade. They exist to trade faster, not to be certain.
- The core edge is index-specific; it was tested and found negative on other instruments.
- At higher risk settings, the single worst historical trade could still brush against a daily loss limit before a soft stop closes it — a known, rare failure mode, not a hidden one.
- Past performance, including everything on this page, does not indicate future results. Markets can enter regimes where any single strategy loses money, including this one.
If a “prop firm EA” page doesn’t have a section that reads like the one above, ask why.
FAQ
No, and you should be skeptical of anyone who says otherwise. What a well-tested EA can give you is a measured, per-attempt probability based on historical simulation — not a guarantee.
The MetaTrader Strategy Tester build runs with no license key and no signup, on your own machine, against your own broker’s historical data. Demo forward testing is also free and unlimited — open a free demo at FxPro, AMarkets or xChief, which takes two minutes and needs no verification. Only live and prop-firm accounts require a paid license.
A single historical backtest is one sample. Monte Carlo resampling generates thousands of alternate, statistically plausible versions of that history to estimate the worst-case tail — the 95th-percentile drawdown — rather than just the one drawdown that happened to occur.
It’s a correction for how many strategy variations were tested before one was chosen. Test enough variations and a good-looking result will appear by chance alone. The Deflated Sharpe Ratio raises the bar a strategy’s final out-of-sample performance must clear based on how many candidates were tried — most retail EA vendors never disclose that count, let alone correct for it.
No. Every leg is long-only, one position at a time, hard stop-loss on every trade. Grid and martingale systems are common in the retail EA space precisely because they can manufacture a smooth-looking equity curve right up until they don’t — most prop firms ban them outright.
It’s built against the standard FTMO-style rule set (daily loss limit, total loss limit, profit target, minimum trading days, best-day consistency where applicable). For the full rule breakdown and how to stay compliant with other firms’ variations, see Prop Firm Rules, in Plain Terms.
Yes, for a live challenge attempt — the EA needs to run continuously to catch every signal and manage open risk. You can find a VPS from out list of the best forex VPS.