Backtesting Gold Strategies in MT5: Spread and Tick Data

You’ve built what looks like a solid gold EA. The backtest curve is smooth, the profit factor is respectable, and you’re ready to go live. Then you switch from “1 Minute OHLC” to “Every tick based on real ticks,” and the equity curve collapses. Trades that looked like winners turn into losers. The spread widens, slippage appears, and suddenly your XAUUSD strategy tester results look nothing like what you optimized.

This isn’t a bug. It’s the most common reality check in gold backtesting. Two settings in MT5 silently decide whether your backtest reflects something that could actually happen on a live account—or a fantasy built on ideal execution. Here’s what actually matters when you backtest gold in MT5, and how to configure both settings correctly for XAUUSD.

Why does my gold EA backtest look so different from live results?

Because the Strategy Tester doesn’t trade on abstract charts. It trades on the prices, spreads, and tick sequences your broker provides—or that MT5 generates when your broker’s data is thin. For gold, that gap between simulation and reality is unusually wide.

Two settings in the Strategy Tester control almost everything about how realistic your backtest will be: the spread mode and the tick model. Most traders adjust one and ignore the other. Or worse, they leave both at defaults that make gold strategies look far more robust than they are.

The first thing to understand is that XAUUSD doesn’t behave like EURUSD. Gold’s spread isn’t constant. It widens during news releases, session transitions, and low-liquidity hours. A strategy tested with a flat 20-point spread on gold is a strategy tested in conditions that may never exist when it matters. The second thing is that MT5’s tick generation modes don’t just affect speed. They determine whether your EA sees price movement between minute bars—or whether those movements simply don’t exist in the simulation.

Get these wrong, and your gold EA backtest isn’t a prediction. It’s a counterfactual.

How should I configure spread for XAUUSD in the Strategy Tester?

Use your broker’s actual spread data, not a fixed number you picked because it made the results look good. Here’s the practical breakdown.

If you’re testing with “Every tick based on real ticks,” the spread comes from the recorded Ask and Bid prices. You don’t get to set it manually—the data determines it. This is the most realistic option when you have high-quality tick history from your broker. The spread widens when it widened historically, narrows when it narrowed. Your EA experiences the same cost environment it would have faced live.

If you’re testing with any other modeling mode, MT5 uses the spread defined in the symbol’s contract specifications or the tester settings. You can change it, and you should. The default “floating” spread in generated modes derives from the underlying M1 bar data, but it’s still a simplified model. For gold, a fixed spread of 20–30 points is a common starting point, but it’s an assumption, not a fact. If your broker’s typical spread during London–NY overlap is 35 points and you’re testing at 15, your backtest is subsidizing every trade.

The real test isn’t finding the “right” spread. It’s understanding how sensitive your EA is to spread changes. Run the same backtest with your broker’s average spread, then run it again with the spread widened by 10–15 points. If the strategy flips from profitable to unprofitable, you’ve learned something critical: your gold EA’s edge is thin enough that execution costs can erase it. That’s not a backtest problem. That’s a strategy problem worth knowing about before you fund an account.

One more thing: check your broker’s gold symbol specifications. XAUUSD and GOLD symbols can differ in tick size, contract size, and minimum stop distance across brokers. A strategy that works on one broker’s gold symbol may behave differently on another’s.

What tick data quality do I actually need for a gold backtest?

Use the highest-quality tick data available: “Every tick based on real ticks.” For gold, the difference between this and every other mode isn’t incremental. It’s structural.

“1 Minute OHLC” generates only four ticks per minute—Open, High, Low, Close—and everything between them doesn’t exist. For a strategy that evaluates on bar close and uses wide stops, that might be acceptable. For anything that reacts to intrabar price movement, it’s dangerous. A trailing stop, a pending order near the market, a breakeven move, a tight take-profit: all of these depend on price paths that OHLC mode simply discards. An EA tested on OHLC might show a winning trade where price touched the stop-loss level before reaching the take-profit—but the simulation never saw the stop because it only checked four prices.

“Every tick” without real ticks generates synthetic ticks from one-minute bars using an interpolation pattern. The path inside each minute is invented. It’s smoother than reality, which means it’s friendlier to your EA than the live market ever was. Generated ticks routinely miss price touches that real ticks hit. Pending orders that would have triggered live may never activate in the test. The “which was hit first, SL or TP” question gets resolved in your EA’s favor more often than it should.

Real ticks replay the Ask and Bid movements recorded by your broker. They include the gaps, the bursts, the moments when gold moved $5 in seconds and the spread doubled. This is the closest you can get to what your EA would have actually experienced. It’s also slower to test and requires more disk space—tick data is stored in TKC files by month and can take a long time to download on the first run. That cost is worth paying for gold.

The modeling mode is not a minor setting you click past. It’s the foundation of everything the backtest claims. A result that doesn’t state which tick model produced it isn’t a reproducible result—it’s a number without context.

Does tick data quality actually change my gold EA backtest results?

Yes, often dramatically. This is the part that catches most traders off guard.

When you switch a gold EA from generated ticks to real ticks, three things tend to happen. First, intrabar stop-outs increase. Price paths that looked smooth become jagged, and levels that seemed safe get hit. Second, pending orders behave differently. A buy limit that triggered in the generated model might never activate in real ticks because the actual price path didn’t touch it—or vice versa. Third, spread costs become real. Generated ticks often use a simplified spread. Real ticks carry the spread that existed at that moment, which for gold can mean 20 points during quiet hours and 80 points during a news spike.

The net effect is that strategies which looked profitable on “1 Minute OHLC” or generated ticks often fall apart on real ticks. This isn’t because real ticks are “wrong.” It’s because the other modes were too forgiving. They hid execution risk that exists in live trading.

A backtest with real ticks on XAUUSD is still a simulation. It cannot perfectly reproduce slippage, latency, or the exact fill you would have gotten from your broker. But it’s the closest available approximation. Anything less accurate is not a conservative estimate—it’s an optimistic one.

What should I check before trusting a gold EA backtest?

Before you move from backtest to demo to live, verify three things about your test configuration. First, confirm the tick model is stated explicitly. If you’re looking at someone else’s equity curve—a vendor, a forum post, an MQL5 Market listing—and it doesn’t say “Every tick based on real ticks,” you don’t know what you’re looking at. Second, confirm the spread reflects your broker’s actual conditions. If the test used a fixed spread that’s lower than what you see in live Market Watch, the results are not comparable. Third, confirm the test period includes different market regimes. Gold’s behavior in a trending month is different from its behavior in a ranging month. A backtest that only covers one regime isn’t a robust evaluation.

And when you eventually run the strategy on a live or demo account, you’ll want to see the actual fills as they happen—not just the backtest curve. That’s where a tool like Camovia Tray fits into a gold trader’s workflow. Once your XAUUSD EA is running on MT5, Camovia Tray keeps your open positions visible from the system tray: hover to check live gold quotes and current P&L without opening the terminal, and close positions directly from the tray popup when you need to intervene[citation:file content]. It doesn’t change how your EA trades or replace the Strategy Tester. It just makes the live side less blind. You can see whether the fills you’re getting match the assumptions you tested with—and act on them without alt-tabbing through five windows.

Backtesting gold in MT5 isn’t about finding settings that make your strategy look good. It’s about finding settings that make your strategy look real. Spread mode and tick data quality are the two levers that control how close your simulation gets to the market. Set both with your broker’s actual conditions, and you’ll know whether your gold EA has an edge worth trading—or just a curve that looked nice in a friendlier world.