Savanna Gold Trend
- Experts
-
Tshivhidzo Moss Mbedzi
I build MetaTrader 5 Expert Advisors for gold and forex, and I publish the parts most sellers hide: the losing months, the strategies I rejected, and a test year I never tune on.
How I build:
• A stop-loss on every trade, from entry. No martingale, no grid, no averaging, no recovery mode. - Version: 1.0
- Activations: 11
Savanna Gold Trend - XAUUSD Expert Advisor for MetaTrader 5
Trend-continuation trading robot for gold on MT5. Pyramids into a running move, never averages down, stop loss on every position.
Savanna Gold Trend is an automated trading robot for XAUUSD on MetaTrader 5 (MT5). It waits for price to break out of its recent range in the direction of the higher timeframe trend, opens a position, and then adds further positions only while that move keeps running in profit. Each position is then managed on its own until its trail takes it out.
The name comes from how it behaves. It sits still for long stretches, commits when a move is genuinely under way, and then stays with that move rather than taking a few points and leaving.
What it does not do:
No martingale. Lot size never increases after a loss.
No grid. No position is ever added against the market.
No averaging down and no recovery mode.
No unprotected trades. Every position gets a stop loss as soon as it is filled.
Every additional entry requires the move to have travelled further in your favour first, so adding into a losing position is impossible by design.
How it works
Trend filter. The EA trades only in the direction of the H1 EMA trend, and only while that trend is still turning the same way.
Entry. A breakout of the recent M5 range by a set distance, with a volatility filter that stands aside when the market is too quiet or too disorderly.
Pyramid. Up to four positions per move. Each additional position requires price to have extended a further set distance in profit, and all of them must fill within a short window of the first. Adding into a losing move is not possible by design.
Exit. Each position trails a share of its own run-up rather than a fixed distance. A large move therefore keeps a large part of itself, while a small one is not given away. The stop only ever tightens.
Risk control. A stop loss on every position placed from its real fill price, a hard limit on separate entries per day, a spread filter, three configurable session windows, an optional daily loss cut-off, and an optional maximum holding time.
Walk-forward test
Backtest settings can be fitted to the data they were chosen on, so a separate check was run. Every parameter was selected using January to April 2026 data only, and then applied without any change to May, June and July 2026, which were not used during that selection.
Result on the unseen period, XAUUSD, 100% real tick data, 1615 USD starting deposit, 1:500 leverage, fixed 0.01 lot:
105 trades, 78.1% won, profit factor 2.29, net result 468.62 USD, maximum drawdown 7.95%, average win 10.11 USD, average loss 15.68 USD, largest single loss 20.84 USD.
These are MetaTrader 5 Strategy Tester results. They are not results from a real account.
Requirements
Symbol: XAUUSD, which is what the settings are calibrated for.
Timeframe: any, because the logic reads M5 and H1 data internally.
Account type: hedging. The EA holds several positions in the same direction, and a netting account merges them, which breaks the per-position exits.
Broker: ECN or Raw spread is preferred. The EA skips entries when the spread is too wide.
Minimum deposit: 1000 USD at 0.01 lot is comfortable. The stop loss is wide by design because the strategy holds through normal pullbacks.
A VPS is recommended.
The EA rescales its price-based settings to whatever symbol it is attached to and will run elsewhere without errors, but the settings are calibrated for gold and no result is claimed on other instruments.
Setup
Attach the EA to an XAUUSD chart on a hedging account. Any timeframe is fine. Keep the default settings, which are the tested configuration. Compare your broker server time against the three session windows and adjust them if the times do not match. Run the EA on a demo account first.
Input parameters
Money and sizing
InpFixedLot - Lot per position. Default 0.01. Set to 0 to size by risk instead.
InpRiskPercent - Percent of balance risked per position when InpFixedLot is 0.
InpMagic - Magic number. Change it if another EA runs on the same account.
InpDeviationPoints - Maximum accepted slippage on market orders, in points.
Symbol adaptation
InpAutoScaleToSymbol - Rescale the price-based settings to the attached symbol. Keep enabled.
InpCalibPrice - Reference price the defaults were calibrated at. Default 3500.
Trend filter
InpTrendTF - Higher timeframe used for the trend. Default H1.
InpEmaFast - Fast EMA period on that timeframe.
InpEmaSlow - Slow EMA period on that timeframe.
InpRequireSlope - Also require the fast EMA to be turning the same way.
Entry
InpBreakoutBars - Number of M5 bars forming the range that price must clear.
InpBreakoutBuffer - How far beyond the range price must trade to trigger an entry.
InpAtrPeriod - ATR period used by the volatility filters.
InpMinATR - Skip trading when volatility is below this.
InpMaxATR - Skip trading when volatility is above this.
Pyramid
InpMaxLegs - Maximum positions per move. Default 4.
InpAddStepPrice - How far price must extend in profit before another position is added.
InpAddWindowSec - All positions of a move must open within this many seconds of the first.
InpAddMinGapSec - Minimum seconds between positions.
InpAddOnlyInProfit - Only add while the move is showing a profit.
Exit
InpTakeProfitPrice - Fixed take profit per position. 0 uses the trail only.
InpTrailStartPrice - Profit at which the trail becomes active. Default 8.00.
InpTrailGiveBackPct - Share of the run-up the trail gives back. Default 70.
InpTrailMinDist - The trail is never tighter than this.
InpUseBreakEven - Optional break-even stop. Disabled by default because it caps winners.
InpBreakEvenAt - Profit that triggers the break-even move.
InpBreakEvenLock - Profit the break-even stop keeps.
InpMaxHoldHours - Close a position after this many hours regardless. 0 disables.
Protective stop
InpSL_ATR - Stop loss as a volatility multiple. 0 uses the fixed distance below.
InpFixedSLPrice - Fixed stop loss distance per position. Default 15.00.
InpMinSLPrice - Lower bound on the stop distance.
InpMaxSLPrice - Upper bound on the stop distance.
Sessions, broker server time
InpUseSessions - Restrict trading to the windows below.
InpSess1Start - First window start hour. Default 1.
InpSess1End - First window end hour. Default 6.
InpSess2Start - Second window start hour. Default 15.
InpSess2End - Second window end hour. Default 19.
InpUseSess3 - Enable the third window.
InpSess3Start - Third window start hour. Default 22.
InpSess3End - Third window end hour. Default 23.
InpTradeFriday - Enable or disable Friday.
InpFridayCutoffHour - Stop opening trades on Friday after this hour.
Risk guards
InpMaxSpreadPoints - Skip entries when the spread is wider than this.
InpMaxMovesPerDay - Maximum separate entries per trading day. Default 3.
InpMaxDailyLossUSD - Stop trading for the day after this loss. 0 disables.
InpRearmSec - Seconds to wait after a move closes before looking for a new one.
Other
InpVerboseLog - Write detailed entries to the Experts log, useful when reporting an issue.
InpComment - Comment attached to each order.
Frequently asked questions
Which account type do I need? A hedging MetaTrader 5 account. This is not optional: the EA holds several positions in the same direction and a netting account would merge them.
Does it work on other symbols? It runs on any symbol without errors because it rescales its settings, but the supplied settings are calibrated for gold only.
How many trades will I see? Up to three separate moves per day with the default settings, each of which may open up to four positions. Often fewer.
Does it use martingale or a grid? No. Lot size is fixed, and a position is only ever added after the move has travelled further in profit.
Why is the stop loss so wide? The strategy holds through normal pullbacks rather than scalping. A tight stop removes the winners along with the losers.
Please read
Past performance does not indicate future results. The testing above covers seven months on one instrument, with three months held aside as an out-of-sample check. All figures are Strategy Tester results, not real account results, and none of them is a forecast. Use money you can afford to lose and test on a demo account first.
