Sarveon

Sarveon is an expert advisor for XAUUSD on MetaTrader 5. Nine independent strategies run side by side on one chart. Each has its own magic number, its own entry timeframe and its own set of distances, and none of them ever adds to a position opened by another.

Read the Honest Limits section before you buy. It is the part that decides whether this expert suits you.

How It Trades

Each strategy scans the recent bars of its own entry timeframe for a swing level: a high that no later bar has exceeded, or a low that no later bar has broken. The level must sit at a minimum distance from the current price, otherwise it is ignored.

The entry is a pending stop order placed at that level with a fixed offset in dollars. A buy stop sits below a resistance level rather than above it, so the order is filled while price is still approaching the level, not after it has been broken. Stop loss and take profit are attached to the order when it is placed and are measured from the order price, not from the fill price.

Each strategy caps how many pending orders and how many open positions it may hold at once, and refuses to place a new order too close to one of its own orders already waiting. Pending orders have a lifetime of their own, from several hours to several days depending on the strategy, and are removed when it expires.

Nine Strategies In One Product

The nine strategies differ in entry timeframe, in how a swing level is confirmed, in how far the order sits from the level, and in the size of the stop and the target. Entries are read from the daily, four hour and one hour charts; exits and position management are handled on lower timeframes, down to M1. They are shipped as one product with one set of code and one input group, not as nine separate items.

By default all nine trade. Two other modes are available: eight strategies with the ninth switched off, and an automatic level that enables strategies in stages depending on the drawdown you allow on the account.

How Positions Are Closed

Four mechanisms work on an open position. Each can be switched off.

Confirmation by the first M1 candle. The first one minute candle that closes after the fill is read once: if it closes against the position and beyond the fill price, the position is closed immediately rather than left to the stop.

Break even. Once price has moved far enough in favour of the position, the stop is moved to the fill price plus a small buffer.

Trailing stop. The stop follows price at a fixed distance, and the trigger for it is measured from the order price, so a fill worse than expected does not shift the whole schedule.

Trailing take profit toward the loss side. When price moves against the position, the target is pulled closer to it, so a trade that has turned can still be closed at the target instead of running the full distance to the stop.

Protection Built In

A real stop loss on every position, attached with the order. Not a wide one, not a hidden one.

No grid, no martingale, no averaging. Volume never grows after a loss.

Spread filter. No order is placed while the spread is above the limit you set, five dollars by default.

News pause. The release schedule of the United States employment report is compiled into the product. Entries stop one hundred minutes before a release and resume sixty minutes after it.

Equity guard. When equity falls below the balance recorded at start by the percentage you set, twenty five percent by default, every position is closed and trading stops. The start balance is stored on disk and survives a restart of the terminal, so the guard keeps measuring from where it started rather than from the restart.

Safety net on the stop. If price passes the level where the stop should have fired and the position is somehow still open, it is closed at market.

Price scaling. Every price distance is multiplied by yesterday's daily open divided by the reference price of the strategy. Settings tuned when gold traded lower do not fall apart when gold trades higher.

What The Tests Show

Strategy Tester, MetaTrader 5, feed of a live broker, 100% real ticks, 44 085 819 ticks over 3231 bars. The window is 2026.02.12 - 2026.08.31, which is 6.7 months, and it lies outside the data the parameters were tuned on. Starting deposit 10000 USD, fixed volume 0.01 lots, all nine strategies enabled.

Net result 3041.93 USD. Gross positive divided by gross negative: 1.98. Recovery factor 4.82. Sharpe ratio 10.07.

863 trades, of which 650 closed positive, 75.32 percent. Average trade 3.52 USD, average positive trade 9.46 USD, average negative trade -14.47 USD. The largest positive trade was 124.49 USD and the largest negative one -107.42 USD: negative trades are individually bigger than positive ones, and the result comes from their number, not from their size.

Maximal balance drawdown 423.38 USD, or 3.27 percent. Maximal equity drawdown 631.06 USD, or 5.17 percent.

These are Strategy Tester figures on historical data. They are not the results of real trading, and they say nothing about what happens next.

Live Testing

The expert has been running on a demo account since September 2026, on a virtual server, with all nine strategies and fixed volume. That is a demo account, not a real one, and the run is short. There is no live track record yet, and nothing here is presented as one.

Requirements

MetaTrader 5, hedging account, symbol XAUUSD. On any other symbol the expert still starts and writes a warning to the journal: it will rescale its distances to that symbol's price, but every setting here was measured on gold, and nothing in this listing describes how it behaves elsewhere.

A broker whose gold spread stays inside the spread filter for most of the session. Entries are pending stop orders, so the broker's stop level matters as well.

A VPS or a machine that stays on: the expert manages open positions and pending orders continuously, and pending orders expire on a schedule.

An account large enough for the volume you choose. With the shipped fixed volume the tested equity drawdown was 5.17 percent of a 10000 USD deposit.

Input Parameters

Srv_MagicBase - base magic number; each strategy adds its own index Srv_Set - which strategies trade: all nine, eight without the ninth, or by account level Srv_FixedLot - fixed volume; zero switches to volume derived from balance and allowed drawdown Srv_MaxAllowedDDPct - drawdown you allow, used by the volume formula and by the automatic level Srv_MaxSpreadUsd - spread above which no order is placed Srv_UsePriceRatio - scale price distances by yesterday's daily open Srv_UseNfpFilter, Srv_NfpBeforeMin, Srv_NfpAfterMin - news pause and its window Srv_UseFakeout - confirmation by the first M1 candle after the fill Srv_UseBreakEven - move the stop to break even Srv_UseTrailSl - trailing stop Srv_UseTrailTp - trailing take profit toward the loss side Srv_HardStopGuard - close at market if price passed the stop level Srv_EquityStopPct - equity drawdown at which everything is closed and trading stops Srv_LedgerEnabled, Srv_LedgerPrefix - write a CSV ledger of orders and fills Srv_ShowPanel, Srv_PanelRussian, Srv_PanelX, Srv_PanelY - on-chart dashboard and its position Srv_DebugLog - verbose log

Honest Limits

A backtest measures the past. Nothing in it makes the next month resemble the tested one.

The tested window is 6.7 months long. That is a short window, and it is stated as a number rather than hidden behind a chart.

The parameters were tuned on other data, and the window above was kept out of that tuning. It is still one window, on one symbol, from one broker's feed.

One of the seven months in the tested window closed negative. The report's monthly chart shows it; it is not cropped out of the screenshots either.

Negative trades are larger than positive ones. The result depends on the share of trades that close positive staying high. When it falls, the arithmetic turns quickly.

The expert has no real-money track record. A demo run of a few days is all that exists today.

Results depend on the broker: spread, stop level, execution and the depth of the gold history in your terminal all change what you get.

Support

Questions and error reports are handled through the comments section of this product or through mql5.com private messages.