Most traders backtest wrong
Your EA looks bulletproof in the backtester. 67% win rate. $150K profit on $10K initial. Then you go live and it gets stopped out on the first trade.
This happens because DIY backtesting and professional backtesting are not the same thing. Professional backtesting shows you what your EA will actually do. DIY backtesting shows you what you want to believe.
Here's the thing: backtesting is where bad trading strategies die. Or hide.
The backtesting gap between pros and DIY traders
Professional backtesting accounts for six things DIY traders ignore:
- Slippage on entry and exit — real brokers don't fill at your exact price. Most traders don't model this. Pros add 2-5 pips slippage on every trade.
- Bid-ask spread — forex has a spread you're not seeing in your backtest. It eats 1-2 pips per round turn. Account for it or your results are fiction.
- Commission on opening and closing — if you're on a regulated US broker like Interactive Brokers or Tastytrade, you pay commissions. DIY backtests ignore this. A 100-trade EA pays $200-$500 in commissions.
- Drawdown during optimization — you're optimizing parameters to fit historical data. This is called overfitting. Professionals use walk-forward optimization to test new parameters on data the EA has never seen.
- Market regime changes — volatility spikes, correlations flip, trending becomes choppy. Professional backtests run through multiple environments. DIY backtests often test one bull market.
- Survivor bias — you're testing on data that survived to today. Historical data doesn't include pairs that delisted or brokers that failed. Professionals account for this in their risk estimates.
Skip any one of these and your backtest is a lie.
What real MT5 backtesting shows
When we backtest for clients, we run three separate models:
In-sample backtest: Test your strategy on historical data it's optimized for. This is your ceiling—the best-case scenario. DIY traders stop here and think they're done. We're just getting started.
Out-of-sample backtest: Test the optimized parameters on data the EA has never touched. This is the floor—the realistic scenario. The gap between ceiling and floor tells you how much you overfitted. If the gap is 50%+ of profit, the EA won't hold up live.
Forward test: Let the EA run live on a demo account for 30-60 days before risking real money. This catches edge cases your backtest missed—liquidity gaps, news spikes, broker requotes that spike spreads from 2 pips to 20 pips in milliseconds.
The traders who follow this process see results that actually hold up live. The traders who skip it blow up accounts.
Why professional results are so different
DIY backtesting optimizes for past performance. Professional backtesting measures predictive power.
When you tweak your EA parameters until the backtest looks perfect, you're not finding a signal. You're fitting noise. Your EA has memorized price history. It will fail the moment conditions change.
Professional backtesting adds friction—slippage, spreads, commissions—specifically to strip away noise. The fewer trades that survive this friction, the more real the remaining edge is.
This is why professional MT5 EAs from Alorny (starting at $100 for simple strategies) come with a full backtest report showing walk-forward, out-of-sample results. You're not paying for code. You're paying for evidence the strategy actually works.
The real cost of DIY backtesting
Let's calculate what DIY backtesting actually costs:
- 40 hours learning MT5, optimization settings, backtest interpretation. That's $1,600 in lost time at $40/hour (US trader average opportunity cost).
- Backtesting 5 ideas before finding one that looks good. Four fail live. Another 30 hours blown.
- Going live with two ideas. One EA loses $800 on live capital before you realize it was curve-fitted. The other crawls. You make $60 in 3 months.
- Total real cost: $3,000 in lost time + $800 in drawdown + 6 months of opportunity cost lost.
A professional MT5 backtest from Alorny costs $100-$300. You get a working demo in 45 minutes. Full delivery in 2-4 hours. The EA comes with a complete backtest report and 30 days of support.
The payoff: an EA that doesn't lie to you about what it can do live.
The framework for professional backtesting
If you want to backtest like a professional, follow this structure:
- Define your market regime. Are you testing in trending markets, choppy markets, or both? Professional EAs work across regimes. DIY EAs work in one.
- Set slippage to 2-5 pips. This is the bid-ask bounce your EA faces on entry and exit. Real brokers like Interactive Brokers model this accurately. Don't skip it.
- Add commission. Calculate actual commission on round-turn trades. Even 0.5% of trade size adds up fast on high-frequency EAs.
- Optimize on one window, test on another. Split historical data in half. Optimize on the first half. Test parameters on the second half. If results drop 50%+, you overfitted.
- Run a forward test on demo for 30-60 days. Let your optimized EA run live without real money. This catches overfitting and real-world conditions your backtest missed—requotes, slippage spikes, news events.
- Use a risk model. Your backtest should show: average trade, max consecutive losses, worst drawdown, Sharpe ratio, profit factor. These numbers tell you if the EA is sustainable.
The mistakes that kill most DIY backtests
Testing only recent data. If you backtest on 2024-2026 data, you miss the 2022 crypto crash and the 2020 COVID spike. You're optimizing for one market environment. It will fail when conditions change.
Ignoring drawdown in favor of win rate. You focus on win rate and total profit. You ignore the $12K drawdown. When you go live with a $10K account, you blow up on trade 3.
Optimizing too many parameters. More parameters means more curve-fitting. A professional EA has 3-5 optimization targets. A DIY EA has 15+. Guess which one works live? The professional one, every time.
Not accounting for broker spread. Your backtest uses tight spreads. Your actual broker uses 2-3 pip spreads. Your winning strategy becomes a losing strategy the moment you connect real capital.
Assuming past returns predict future returns. Backtests don't forecast the future. They show what worked. Professional traders treat backtests as a minimum bar, not a prediction of tomorrow's P&L.
How to spot a real backtest vs. a fake one
Real backtests show:
- Walk-forward out-of-sample results (tested on data the EA never saw)
- Slippage and spread accounted for (not just profit)
- Maximum drawdown and consecutive losing trades displayed
- Profit factor and Sharpe ratio (not just win rate)
- Multiple market regimes (not just one trending period)
- Commission and broker costs factored in
Fake backtests (DIY trader backtests) show:
- Perfect 75%+ win rate
- No slippage or spread modeled
- Tested only on trending markets
- Only total profit, zero risk metrics
- Optimized on the exact data being tested
If a backtest looks too good to be true, it is.
Frequently asked questions
- Is backtesting an Expert Advisor legal in the United States?
Yes. Backtesting your own EA is legal and encouraged. The CFTC and NFA have no rules against testing strategies on historical data for personal trading. What matters is that you disclose results honestly (no fabricated statistics) if you're selling the EA to other traders. If you're backtesting for your own account, you have full freedom to test and iterate. - Which US brokers show the most accurate backtest results in MT5?
Interactive Brokers, Tastytrade, and TD Ameritrade (via thinkorswim) offer the best MT5 data quality. Interactive Brokers especially models slippage and spread accurately on historical data. When you backtest on their data, your live results will be closer to your backtest results because the broker conditions are similar. - How long should I forward test on demo before going live with real money?
Minimum 30-60 days on a demo account. This gives your EA time to experience multiple market conditions—trending days, choppy days, news events, gap openings. If your EA survives 60 days of demo without unexpected drawdowns or requotes, you can deploy with real capital. If it blows the demo account, something is wrong with the backtest. - Why do some EAs have 85% win rates in the backtest but lose money live?
Win rate and profitability are not the same thing. You can win 85% of trades and still lose money if your average loss is bigger than your average win. That's a poor profit factor. Professional backtests show both: win rate AND profit factor. A 60% win rate with a 2.0 profit factor beats an 85% win rate with a 0.8 profit factor every time. - Can I use the same backtest results on different brokers?
No. Each broker has different spreads, slippage, and historical data quality. An EA that's profitable on Interactive Brokers might lose on a retail broker with 3-pip spreads. Always test on the actual broker you plan to trade with, or accept a margin of safety in your backtest results.
The truth about backtesting
Professional backtesting is expensive because it's honest. It strips away every excuse—slippage, spread, drawdown, overfitting. It shows you what your EA will actually do, not what you hope it will do.
DIY backtesting is cheap because it's forgiving. You optimize until the results look perfect. You ignore spread. You test only on winning markets. You convince yourself you have an edge.
The traders who hire professionals to backtest—even at $300-$500 per EA—go live with strategies that work. The traders who DIY end up trading demo accounts forever, or blowing up real accounts, or both.
At Alorny, we've completed 660+ EA projects with full backtesting reports showing walk-forward results, slippage, spread, drawdown, and risk metrics. We don't hide behind optimized numbers. Your backtest determines whether your next 12 months is profitable or painful.