Gold Night Drift (MT5): my setup guide, server-clock checks and how to read the IS/OOS record

I develop Gold Night Drift at D7 Bots for traders who want a defined approach to overnight gold exposure and accept variable yearly results. It trades XAUUSD with long exposure during the night. This guide covers the setup I ask users to check, the importance of the broker's server clock, the messages to inspect on the first run, and the positive and negative parts of the published record.

What the EA does

Gold Night Drift uses a proprietary time-based approach, with one strategy position at a time, risk-based position sizing and protective orders. It does not use a grid, martingale or a loss-recovery basket. I keep the strategy at its tested standard while checking that the trading environment is compatible.

The trading instrument is selected explicitly. Moving the EA to another chart does not automatically change its target instrument. It receives callbacks from the host chart and checks the configured gold instrument's data and trading conditions. A chart attachment and an accepted trade are different stages of operation.

Installation and the first run

For a first online run of a licensed copy, I use an existing demo environment and work through these checks before leaving the terminal running.

1. Locate the installed product. In Navigator, find Gold Night Drift under Expert Advisors. Confirm the installed product and version rather than reusing an unrelated gold EA's saved setup.
2. Check the exact instrument. In Market Watch, identify the broker's gold symbol, including any suffix or prefix. Match the EA's trading-symbol selection to that name. Inspect the specification, load its history and confirm that quotes update.
3. Use the reference chart and risk. Attach the EA to the broker's XAUUSD instrument on M30. Risk per operation is adjustable; the published evidence uses 1.00% risk per operation. Leave the remaining strategy settings at the tested standard.
4. Verify the server clock before enabling trading. Confirm the broker server's UTC offset and daylight-saving convention with the broker. The computer's local time is not a substitute. Check compatibility with the release's clock configuration, including seasonal changes. This is an installation requirement, not a reason to experiment with the strategy's trading times.
5. Check trading permissions. Enable Algo Trading and the EA's algorithmic-trading permission in its properties for the demo run. Confirm that MT5 is connected and the host chart receives ticks.
6. Read Experts before assuming it is ready. Open Experts in the Toolbox. The GoldNightDrift startup entry contains initialized and reports the host, target and dataReady status. Check the instrument and chart. If it says target XAUUSD data are not ready, initialization has not established data readiness: the EA retries on host callbacks without substituting another symbol or feed.

I retain the installed version, broker symbol, clock convention, chart, test dates and relevant log entries. If startup reports timezone self-test failed, I resolve that failure before proceeding. A successful built-in clock check does not independently establish the broker's actual time convention.

What I check in Experts when nothing happens

Operational rejections use the prefix GoldNightDrift: reject, followed by a stage and code. CONVERSION_FAILED identifies a failed time conversion; INVALID_OR_STALE_TARGET_TICK identifies a target quote that failed the executable-quote check; SPREAD_GUARD identifies a blocked spread check. I investigate the corresponding broker data or clock conditions rather than change the strategy to force an entry.

INVALID_VOLUME means sizing did not produce an acceptable volume. ORDER_REJECTED means the order attempt was not accepted. FOREIGN_POSITION identifies a position on the target symbol belonging to another trading identity. Keep the surrounding log for support.

Some waiting paths return silently, including a closed target session or unavailable synchronized data. The EA also checks quote validity against the target bar and server time. A quiet Experts tab needs to be read alongside connection status, quotes and history; it does not establish that an entry was due.

Reading the published IS/OOS record

I start with the reference environment: MetaQuotes-Demo, model 1 (M1 OHLC), XAUUSD M30 and a virtual initial deposit of USD 10,000 for each test. I then read the two windows separately.

• IS: 1 January 2020 to 31 December 2024. Net profit USD +2,653.26; 1,012 trades; profit factor 1.25; maximum balance drawdown 7.03%.
• OOS: 1 January 2025 to 3 October 2026. Net profit USD +856.76; 238 trades; profit factor 1.28; maximum balance drawdown 5.54%.

Each window comes from its own independent Strategy Tester run. The gallery joins the closed-deal balance curves with a marked boundary and offsets the OOS balance for display after its deposit restart. Metrics remain those of each original run. The joined chart is neither a continuous account replay nor an equity curve showing floating losses.

Does the positive OOS hide a bad year? It can if I look only at the aggregate. The page reports USD +1,164.79 in 2025 and USD -308.03 in partial 2026 through 3 October. The IS also included USD -317.97 in 2021. The profitable OOS total therefore includes a losing later portion, not an uninterrupted run of gains. I keep that breakdown beside the aggregate when assessing whether this overnight exposure suits the intended use. The partial year is not a complete annual observation.

Limits of this record

These are historical Strategy Tester simulations, not live-account results. Overnight exposure entails financing and gap risks. Higher risk settings accepted by the EA, up to 10.00%, were not measured in the published evidence and can materially increase loss and drawdown. Protective orders do not guarantee an execution price.

Reference ledger costs were IS swap USD -536.92 and OOS swap USD -61.16, with commission USD 0.00 in each window. These costs are already included in net results and are not a broker-independent cost model. A separate fee column is unavailable. Contract size, tick value, volume step, margin, spread and trading sessions require verification. Other host charts can change execution timing and results. Historical simulation does not guarantee future performance.

Questions I address before first use

My broker adds a suffix to gold. What should I select? Use its exact symbol in the trading-symbol selection. The EA selects the name supplied; it does not automatically search for a substitute gold instrument. Check the target named in Experts.

Does it require netting or hedging? The inspected release does not restrict initialization to either account mode and uses MT5's account-aware trading functions. I check the intended arrangement in demo rather than treat that as universal broker certification.

Can I use a broker with another server time zone? Not automatically. This release validates a specific clock convention at initialization; it does not discover arbitrary broker conventions. Have compatibility checked before use rather than changing trading times to compensate.

Is a VPS required? The EA requires no particular hosting provider. Its logic needs a running, connected MT5 terminal receiving ticks, including for position management. A VPS is an alternative location for that terminal. After migration, I verify the chart, clock configuration and Experts entries again.

Where does the free demo run? The free Market demo runs in the MT5 Strategy Tester, not on an online chart. A subsequent demo-account chart test requires a licensed copy. Save the broker, symbol, period, configuration and report together.

Can I run it with other EAs? I start with a single instance for gold and avoid other trading activity on that symbol. Duplicate instances with the same trading identity can manage the same position; the foreign-position check does not isolate those duplicates. Other symbols still share account equity and margin.

Gold Night Drift: product documentation and support on MQL5 Market.