Is there a live track record that someone else verifies?
A backtest shows what a strategy would have done. A live account shows what it does. Look for a track record hosted by a third party such as Myfxbook or MQL5 Signals, where the seller can't edit the history. Check whether it is a demo or real-money account, and how long it has been running: a few weeks proves very little; many months across different market conditions proves more.
Red flag: only screenshots, or a "verified" badge with no link you can open.
Does every trade have a stop loss?
Martingale (doubling the size after a loss) and grid systems (adding orders as price moves against you) can produce very smooth equity curves for months, until one strong gold trend wipes the account. Ask directly whether every position has a stop loss from the moment it opens, and watch a few live trades to confirm.
Red flag: many open positions in the same direction at once, or a "recovery" mode after losses.
Read profit factor and drawdown together
Profit factor is gross profit divided by gross loss: above 1 means the system made more than it lost. On its own it says nothing about the pain along the way. The maximum drawdown tells you how far the account fell from a peak. A high profit factor with a huge drawdown, or a tiny number of trades, is fragile.
Red flag: results that show only total profit or win rate, with no drawdown figure.
Check how the backtest was made
In the MetaTrader 5 Strategy Tester, "every tick based on real ticks" with high history quality (close to 99%) is far more realistic than modelled ticks. The test should cover several years of gold, include realistic spreads and commission, and use the same broker conditions you'll trade on.
Red flag: a short test period, "open prices only" modelling, or a test on a different symbol or broker than the one recommended.
Compare live results with the backtest
The real test of a backtest is whether the live account behaves in a similar way: similar trade frequency, similar average win and loss, and drawdowns in the same range. Live results will always be somewhat worse, because of slippage and spreads at news. Large gaps between the two mean the backtest was probably over-fitted.
Red flag: live trading that looks nothing like the backtest that sold you the EA.
Match the broker conditions
Gold strategies that trade often are sensitive to spread, commission and execution speed. An EA tested on a Raw (ECN-style) account may perform very differently on a standard account with wider spreads. Use the account type the EA was tested on.
Red flag: "works on any broker" for a strategy that trades several times a day.
Know who you're dealing with
A credible seller has a name, discloses affiliate links, says clearly when an account is demo, and never promises returns. Real developers don't ask for your trading password and don't contact you first offering to "manage" your account.
Red flag: guaranteed or fixed monthly returns, pressure to pay now, or an "account manager" in your direct messages.
Decide what risk you can live with
Take the largest drawdown in the backtest and assume live trading can be worse. If that loss would hurt you, use a smaller account or lower risk, or don't use an EA at all. Only trade money you can afford to lose.
Red flag: anyone telling you a trading robot is "safe" or "risk-free".