Ecfx Xau Sentinel / Warden R2 v2.20: MT5 Setup and Testing Guide

For MT5 Market version 2.20 (source build 2.020). Guide dated 1 October 2026.

Ecfx Xau Sentinel is the MQL5 Market listing name for the Ecfx XAU Warden R2 project. Check the installed version before copying settings from a video, preset or earlier release.

Official product page and free Strategy Tester demo

1. Install the correct edition

In your intended desktop MetaTrader 5 terminal, sign in to the MQL5 account that owns the product and install Ecfx Xau Sentinel from your Market purchases. Keep licensing and updates within the official Market workflow.

The free Market demo runs in Strategy Tester. It is not a free licence for a running demo-account or real-account chart. Use the tester demo to evaluate the product before purchasing.

2. Prepare a hedging demo account

Use a hedging account and your broker’s actual Gold symbol. The supplied portfolio needs separate positions and strategy ownership. A netting account combines positions on the same symbol and changes that behaviour. The documented source warns on netting; that does not establish support for the intended portfolio on netting accounts.

Start with one instance on an H1 Gold chart. H1 is a practical working layout: the strategies also use their own internal timeframes. Load the exact MT5 .set file for the installed build and save a named copy before making changes.

Review ComboExecutionMode before enabling chart trading. The documented factory mode allows demo and real charts; deliberately select demo-only operation for initial forward checks. Enable terminal and EA trading permissions only when that chart is intended to submit orders.

3. Check the broker and strategy clock

Record the exact symbol, digits, point size, tick size, tick value, contract size, minimum lot, volume step, leverage and trading hours. Check spread, commission and swaps. There is no universal minimum deposit: broker minimum volume and the combined enabled exposure matter.

Engine B session inputs use minutes after midnight in the strategy clock: 120 means 02:00, 420 means 07:00, and 1080 means 18:00. Check PG_InpSessionShiftMinutes against broker-server time.

Engine C uses OB_BrokerUtcOffsetMinutes (server minus UTC). Tester runs use the explicit offset; OB_AutoLiveUtcOffset can derive the offset on a running chart. Engine A has separate GMT and NFP settings. One offset does not configure every engine.

Engine A historical NFP filtering uses an approximate first-Friday schedule unless exact NFPBrokerTimes are supplied. Record exact broker timestamps when testing event exclusions. This is not a universal filter for all high-impact news.

4. Understand the three engines

Engine A uses price-level breakout profiles, pending orders and profile-specific exits. Engine B combines a session breakout and scheduled entries; its additional module C is disabled by default. Engine C combines technical models and additional breakout families.

The engines use distinct magic-number ranges but share balance, equity and available margin. Their profits cannot simply be added to predict a combined portfolio. Several strategies can lose during the same Gold movement.

Check the top-level engine switches, strategy-family switches, sizing basis, lots, spread filters and identifiers. PG_InpEnableA and PG_InpEnableB are modules within Engine B, not the top-level Engine A and Engine B switches. Leave inactive compatibility fields and unsupported legacy families at their supplied values.

5. Review risk controls explicitly

A sizing setting is not an enforced loss ceiling. Engine A’s MaxAllowedDD participates in sizing and selection. In this documented build, EnforceMaxTotalDrawdownGuard defaults to false and PropFirmMaxDailyDD defaults to zero. Entering a drawdown number alone does not activate those protections.

Engine B’s PG_InpCloseTradesOnEquityDrawdown defaults to false. Its percentage threshold does not activate the switch. Account-wide closing must remain disabled in the combined portfolio.

Engine C’s shared equity guards manage its owned exposure, rather than providing a universal close-all rule for Engines A and B. OB_BacktestBypassPermanentMaxDrawdownLock defaults to true: explicitly record it and compare with the bypass disabled when assessing chart-like risk-lock behaviour.

Check OnlyUp and other sizing-reference settings: retaining a prior higher balance or equity can keep sizing elevated after losses. Stops and software guards cannot guarantee an exact maximum loss through gaps, slippage, rejected operations or disconnections.

6. Run a reproducible backtest

1. Use the exact build and saved preset. Record broker, Gold symbol, H1 working timeframe, account model, deposit, currency, leverage and date range.
2. Choose Every tick based on real ticks where appropriate broker history is available. Check data quality, commission, swaps and execution assumptions.
3. Start with a short visual test. Confirm engine initialization, plausible volumes and stop distances, expected session times, and accepted orders. Investigate repeated errors in Experts and Journal.
4. Then test a longer period covering different market conditions. Compare net result after costs, balance drawdown, equity drawdown, trade count, margin usage, losing sequences and recovery time.
5. Test A-only, B-only, C-only and combined configurations separately. Keep some dates out of parameter selection and check sensitivity to costs and nearby settings.
6. Save the completed report, exact preset and logs. A compile, running panel or attractive screenshot does not establish a completed or reproducible test.

Related guide: how to backtest a Gold EA in MT5

7. Forward demo operation

With the purchased edition, observe the intended configuration on a hedging demo account. Confirm actual order acceptance, timing, reconnect behaviour and broker session closures. Keep the terminal connected and the EA running for software-managed trailing, timed exits and virtual expiry.

Do not run instances with overlapping magic-number ranges. Avoid changing identifiers while positions are open. Stopping or removing the EA does not guarantee that positions close; inspect remaining positions and pending orders.

The panel is an operational aid. An enabled or active engine is not a promise that a trade is due. Filters, warmup, cooldowns, market hours and risk locks can legitimately prevent new entries.

8. Troubleshooting and support

No trades: check history, permissions, engine switches, market hours, filters, margin and risk locks. Do not disable every safeguard just to produce activity.

Invalid volume or stops: check minimum volume, lot step, tick size and broker stop-distance requirements. A calculated size can fall below the broker minimum.

Results differ from a video: compare build, preset, dates, data model, costs, symbol specifications, offsets, leverage and drawdown bypasses. Earlier Warden videos may show different builds.

For help, use the product comments or MQL5 private messages. Include the installed version, broker symbol, account type, timeframe, saved preset, test dates, report and relevant Experts or Journal lines. Remove account identifiers, passwords and other credentials.

Risk disclosure: automated leveraged Gold trading can cause substantial losses. Historical simulations and demo results do not guarantee future performance, a fixed drawdown ceiling or prop-firm eligibility. This guide documents software operation and is not personalised financial advice.