EA due diligence — verifying any Expert Advisor before real capital
Apply a 10-point due diligence checklist that separates EAs worth testing from those worth avoiding.
Dernière révision :
EA due diligence in 90 seconds
When MyForexFunds was shut down by US and Canadian regulators in 2023, thousands of traders lost access to funded accounts. Many had bought EAs specifically marketed as 'MFF challenge passers' — EAs with no live track record, no identifiable vendor, and hard-coded risk rules that prevented daily-loss-limit compliance.
- →Three checks that matter most: live track record on a real account, configurable risk parameters, and identifiable vendor.
- →If any of these three critical checks fails, stop — do not proceed regardless of how compelling the marketing is.
- →The remaining 7 checks are amber signals: two or more failures mean extend testing, not abandon entirely.
- →Always run on demo at your broker before live, even if all 10 pass.
The EA market has good products and predatory ones. The 10-point checklist separates them in under an hour.
EA due diligence — 10-point checker
Work through each point. Critical items are marked — any critical failure stops the evaluation.
Myfxbook or FX Blue link showing a real (not demo) account, broker-verified connection, at least 6 months of history.
Account Type = Real, Connection = Verified, history ≥ 6 months
Daily loss limit, drawdown cutoff, and lot-size mode appear as user inputs in the MT5 input panel — not hard-coded.
DailyLossLimit, MaxDrawdown, LotMode all present as inputs
Vendor has a real name, verifiable web presence (company registration, LinkedIn, social history with depth), and responsive support.
Can verify vendor identity independently; support responds within 48h
Strategy Tester report shows 99% modelling quality, realistic spread settings, and equity curve reacts to known market stress events.
Modelling quality ≥ 90%; no suspiciously flat periods during known stress events
Dated version history with plain-English explanation of changes. Updated within last 12 months.
At least one update with a dated note in the past 12 months
Reviews found on third-party platforms (Trustpilot, forex forums) outside the vendor's own site, with substantive technical content.
At least a few independent reviews with specific, verifiable claims
Vendor is not offering to control your capital, copy trades to their account, or requiring you to fund an account they manage.
EA is sold as a tool you operate — no capital control by vendor
EA runs on your specific broker's demo account as described. Execution, spread, and latency match vendor claims.
2-4 weeks of demo operation at your target broker with acceptable performance
One-time purchase or subscription with clear terms for what happens if subscription lapses or vendor closes.
Licensing terms documented; no surprise renewal or account-lock provisions
Pre-sale technical question answered with depth and speed. Post-sale support channel exists and is monitored.
Pre-sale question answered within 48h with technical depth
Score: 0/0 checked (10 total)
The 10-point framework
Critical checks (fail = stop)
Three checks are binary. If any of them fails, no amount of compelling backtest performance or vendor testimonials changes the verdict — you do not proceed to live capital.
Check 1: Live track record on a real account. A verified Myfxbook or FX Blue link showing a real (not demo) account with at least 6 months of live trading history. The account must be broker-verified, not manually uploaded. Check 2: Risk parameters are user-configurable. Daily loss limit, drawdown cutoff, and lot-size mode must appear as inputs in the EA's MT5 panel — not hard-coded. Check 3: Identifiable vendor. The selling entity must have a real name, a verifiable web presence (company registration, LinkedIn profile, social accounts with history), and a support contact that actually replies.
Amber checks (two failures = extend testing)
Seven checks are contextual. One failure is a yellow flag; two or more failures mean run demo for at least 90 days before considering live, not necessarily abandon.
Amber check A: Backtest quality. Does the backtest show realistic spreads and slippage? Are there suspiciously flat no-loss periods that coincide with news events? Was the data quality 99% modelling quality in the Strategy Tester report? A good backtest shows regular drawdown — a perfect equity curve is a curve-fitting signal.
Amber check B: Changelog. Can you see dated version history with plain-English explanation of what changed? An EA with no update history for 12+ months is potentially drifting without monitoring.
Amber check C: Independent reviews. Are there reviews on platforms outside the vendor's site (Trustpilot, forex forums, EA review sites)? Only positive reviews from anonymous accounts is a yellow flag.
Amber check D: Licensing terms. Is this a one-time purchase or a subscription? What happens to your running EA if the vendor closes? Can you use it on multiple accounts or only one?
Amber check E: Demo at your broker. Does the EA perform on your broker's demo environment as described? Slippage and spread conditions vary significantly by broker. Test at your specific broker before live.
Amber check F: No 'managed trading' bundle. Is the vendor offering to 'manage' your capital, copy trades to their account, or requiring you to fund an account they control? If yes, that is no longer an EA sale — it is an unregulated investment scheme.
Amber check G: Support quality. Send a pre-sale technical question (e.g. 'what broker types are compatible?'). Measure the response time and depth. Vendors who don't respond pre-sale almost certainly don't respond post-sale.
Key terms
What the 2023 prop-firm shutdown taught EA buyers
MyForexFunds fermé par la CFTC — 135,000 traders privés d'accès du jour au lendemain — August 2023
The August 2023 shutdown of MyForexFunds by the CFTC and Ontario Securities Commission affected over 135,000 traders. In the aftermath, the EA market that had grown around MFF challenges came under scrutiny. Analysis of popular 'challenge-passing' EAs sold during 2022-2023 found three recurring patterns: (1) hard-coded risk rules preventing the trader from setting a real daily loss limit, (2) no verifiable live track record — only backtest screenshots, and (3) vendors who disappeared from contact within weeks of the shutdown.
The lesson is not 'avoid EAs' — it is 'apply the checklist'. The EAs that held up through the MFF shutdown had verifiable live accounts, user-configurable risk parameters, and vendors who remained reachable and shipped updated configurations for compliant brokers.
SourceRun the checklist
Pick any EA you are currently evaluating and work through the checklist widget below. The widget tracks your responses and produces a pass/warn/fail verdict.
- 1Collect the EA vendor's Myfxbook link, website URL, and MT5 input panel screenshot before starting.
- 2Work through the three critical checks first. If any fail, stop.
- 3Continue with the seven amber checks. Note any failures.
- 4Review the verdict from the widget and decide: proceed to demo, extend testing, or decline.
Mastery check
Mastery check
Testez vos connaissances avec 5 questions. Réussite avec 80/5 bonnes réponses.
Reflect
Réflexion
Notez vos réponses honnêtes — sauvegardées uniquement sur cet appareil. Reprenez-les la semaine prochaine pour repérer des tendances dans votre façon de penser le trading.
Forensic EA analysis
The 10-point checklist is a practitioner's heuristic. The underlying discipline is financial forensics — applied to EA products rather than company accounts.
Forensic backtest analysis
A professional EA analyst reads a backtest like an auditor reads a balance sheet. Key diagnostic: what happened during the known market stress events of the backtest period? If EURUSD backtest covers 2020-2024, does the equity curve show the COVID crash (Mar 2020), the EURUSD parity move (Sep 2022), or the yen carry-trade unwind (Aug 2024)? If the equity curve is flat through all three, the EA either had positions closed manually, the data gaps masked the events, or the parameters were over-fitted on a dataset that excluded them. Real strategies react to real events.
Legal exposure of unregulated managed trading
In most jurisdictions, accepting client capital for trading decisions requires a financial licence (FCA in the UK, ASIC in Australia, NFA/CFTC in the US). Vendors who bundle 'I'll manage it for you' with an EA sale without holding these licences are operating outside regulation. In the MyForexFunds case, the allegation was not just EA fraud but operating an unregistered commodity pool — carrying criminal exposure, not just civil. Traders who participate in these schemes can lose their capital with no regulatory recourse.
Statistical significance of short track records
A 3-month live track record with 50 trades has significant statistical uncertainty. At 60% win rate and 50-trade sample, the 95% confidence interval on win rate spans roughly 46%-74%. You cannot distinguish a 60% edge from a 55% edge or even random noise at this sample size. Six months with 150+ trades begins to give meaningful signal; 12 months with 300+ trades starts to approach conventional statistical significance thresholds. Always ask: how many trades are in this track record, not just how many months.
Sources
Afficher la réponse
1. Verified live track record on a real account (Myfxbook/FX Blue, 6+ months, real not demo). 2. User-configurable risk parameters (daily loss limit, drawdown cutoff, lot mode in MT5 inputs panel). 3. Identifiable vendor (real name, verifiable web presence, responsive support).
Matériel éducatif uniquement — ce ne sont pas des conseils en investissement. Le trading comporte un risque de perte en capital. Entraînez-vous toujours sur un compte démo et utilisez un stop-loss. ← Retour à Automated Trading