Most algorithmic trading services show you an equity curve and ask you to trust it. We've taken a different approach — full transparency on methodology, data quality, test parameters, and results. This is a complete breakdown of how the ForexFloor XAUUSD strategy was built, tested, and validated.
The full MT5 Strategy Tester report is publicly available for download on our results page. Everything in this article can be verified against that document.
Why Backtesting Matters — And Why It's Often Done Wrong
Backtesting is the process of running a trading strategy on historical data to see how it would have performed. Done correctly, it's the closest thing to proof that a strategy has a genuine edge. Done incorrectly, it produces results that look spectacular but mean nothing.
The most common backtesting failures:
- Low-quality data — Using broker data that's been modified, has gaps, or doesn't accurately represent tick-by-tick price movement. Strategies tested on low-quality data have completely different real-world performance.
- Overfitting — Optimising parameters so specifically to historical data that the strategy becomes a perfect description of the past rather than a predictive model for the future.
- Cherry-picking periods — Testing only during favourable market conditions and excluding drawdown periods.
- Ignoring transaction costs — Not accounting for spreads, commissions, and swap rates that erode real-world returns.
- Small sample sizes — Drawing conclusions from 30 trades when statistical significance requires several hundred.
We designed our testing process specifically to avoid every one of these failure modes.
The Testing Environment
Platform
MetaTrader 5 Strategy Tester — the industry standard for MQL5 Expert Advisor backtesting. MT5's Strategy Tester uses variable spread modelling and tick-by-tick simulation, making it significantly more accurate than MT4's bar-based testing.
Broker and Data Source
Vantage International Group Limited, live account environment. The backtest was run on Vantage's historical XAUUSD+ tick data — the same data feed used by live accounts. This eliminates broker feed discrepancies that can make backtests look better than live trading.
Data Quality
99% history quality — the highest possible rating in MT5 Strategy Tester. This rating means the simulation reconstructed every tick between recorded price points using mathematical interpolation, producing a simulation environment that is as close to real market conditions as backtesting can achieve.
Test Period
January 1, 2022 — March 10, 2026. Four years and two months, spanning 49,434 bars on the M30 timeframe and processing 220,872,001 ticks.
Why start in 2022? 2022 was chosen as the start date deliberately. Gold entered a sustained downtrend in 2022 that lasted most of the year — one of the harshest conditions for any gold long-only strategy. If the algorithm couldn't survive 2022, it wasn't worth deploying.
The Strategy: Smart Money Concepts on M30
Timeframe
M30 (30-minute candles). This was selected after testing multiple timeframes. M30 provides enough trading opportunities to produce statistically meaningful sample sizes while avoiding the noise and whipsaw of shorter timeframes like M5 or M15.
Signal Logic
The algorithm is built on Smart Money Concepts (SMC) — a framework for analysing institutional order flow through price structure. Two signal types are used:
- Break of Structure (BOS) — identifies when price breaks a previous significant high, signalling that institutional buying pressure has shifted the market structure bullish
- Change of Character (CHoCH) — identifies when price changes its trending behaviour, signalling a potential reversal or continuation move of higher probability
Trend Filter
An EMA80 filter prevents the algorithm from taking long trades when the medium-term trend is bearish. This was the single most important addition to the strategy — it was the filter that allowed the system to survive 2022's gold downtrend while other long-only strategies were destroyed.
We tested EMA periods from 50 to 200. EMA80 produced the best balance between trend responsiveness and false signal filtering across the full test period — not just the optimisation window.
Risk Management
- Risk per trade: 3% of account equity (the baseline scenario)
- Stop-loss: placed at the structural low of the identified setup
- Take-profit: 5:1 reward-to-risk ratio
- Break-even: stop moved to entry once position reaches 1:1 profit
- No martingale, no position averaging, no grid trading
Full Results: January 2022 — March 2026
Year-by-Year Breakdown
This is where the credibility test really lies — not the overall 4-year number, but how the strategy performed in each individual year.
2022 — The Stress Test
Gold dropped over 20% in a sustained downtrend. The EMA80 filter kept the algorithm largely out of the market during the worst of the decline. The strategy took fewer trades than any other year but preserved capital — ending 2022 approximately flat to slightly positive. Most competing long-only strategies suffered catastrophic losses this year.
2023 — Rebuilding
Gold recovered and began its structural uptrend. The algorithm returned to full participation, identifying high-probability BOS and CHoCH setups as the market regained directional conviction. Moderate drawdown periods in mid-2023 tested the risk management framework before strong Q4 performance.
2024 — Acceleration
Gold broke above $2,000 and began its historic rally toward all-time highs. The algorithm's trend-following logic captured multiple extended moves. Compounding effects began to dominate results as position sizes scaled with the growing account balance.
2025–2026 — Compounding at Scale
Gold continued its bull run past $3,000 and toward $5,000. Position sizes had scaled significantly with the compounding account. Large winning trades in late 2025 and early 2026 drove the bulk of the spectacular final balance figure. This is where percentage-based position sizing creates exponential returns — and also where maximum drawdown figures are at their highest in absolute dollar terms.
What the Numbers Don't Tell You
A 84,277% return sounds extraordinary — and at 3% risk per trade with full compounding it is. But several context points are critical for a realistic understanding:
- The 37.54% drawdown is real. There were extended periods where the account was down over 30% from its peak. Many traders would have stopped the strategy during these periods and locked in losses.
- 3% risk per trade is aggressive. At lower risk settings (0.5% or 1%), the total return is significantly lower but the drawdown is dramatically reduced. The strategy was tested at multiple risk levels.
- Compounding magnifies everything. The final balance figure is the result of reinvesting all profits. Most real accounts are partially withdrawn. Real-world performance will look different.
- Live trading introduces real spreads and commissions. The backtest accounts for Vantage's standard spreads and commissions, but real execution can vary.
Why We're Publishing This
Transparency is our competitive advantage. The copy trading signal industry is full of services that hide behind equity curve screenshots and unprovable claims. We're publishing the full methodology and the actual MT5 report because we believe traders deserve to make decisions based on real evidence.
If you find a flaw in our methodology, we want to know. If the strategy fails in live deployment, we'll say so publicly. The data is the data.
Download the full backtest report
The complete MT5 Strategy Tester report is available on our results page — 783 trades detailed, every entry and exit timestamped.
View Results & Download Report