Not All 99% Backtests Are Equal: How Tick Data Quality Impacts Your Strategy in MetaTrader 5

Backtesting in MetaTrader 5 (MT5) often provides a sense of confidence especially when your strategy shows a high win rate and the tester reports “99% modeling quality.” But here’s what many traders overlook:

✅ Not all 99% backtests are created equal.

The source and quality of your tick data can drastically influence your results—even if MT5 tells you that the backtest was near-perfect.

This guide explores how tick data size and quality differ between brokers, why that matters, and how to validate and compare your testing environment for accuracy and realism.

🔍 What Is "Modeling Quality" in MT5?

MT5 supports real tick data for backtesting, especially under the setting:

When enabled, MT5 attempts to simulate historical trades using actual tick-by-tick data (bid/ask changes, volume, etc.) from your broker's server or a data provider.

While the platform might show 99% modeling quality, this doesn't mean:

• The data has the same granularity across brokers

• The tick frequency is identical

• The spread behavior is realistic

• The volume of market activity is equally represented

📦 Why Tick Data File Size Varies Between Brokers

Tick data file sizes often differ between brokers for several technical and operational reasons:

1. Quote Frequency

• Some brokers record every price change—others throttle updates.

• More frequent updates = more precise price path and larger files.

2. Spread Behavior

• ECN brokers often have floating, dynamic spreads that reflect real-time liquidity.

• Market makers may offer fixed or smoothed spreads.

• Tick data that captures these micro-changes in spread produces richer, larger datasets.

3. Execution Model

• STP/ECN brokers often mirror true market conditions (requotes, slippage, and price spikes).

• This makes backtests more aligned with live results—at the cost of file size.

4. Timezone and Trading Week

• Brokers operating 6 days a week (Sunday open) generate more tick records than those using a 5-day week.

5. Data Compression and Storage Policy

• Some brokers compress or pre-filter ticks to reduce server load or file size.

🧪 Case Study: Broker A vs Broker B

Let’s say you're using a scalping Expert Advisor (EA) that opens trades with a 31-point profit buffer before trailing. You run two backtests on EURUSD M1 with the same EA and get:

Despite identical modeling quality, the larger dataset may include:

• More micro-tick movements

• Realistic widening/narrowing of spreads during news

• More frequent price flickering (which matters for scalping)

👉 Which one is better?

For precision-critical strategies like scalping, Broker B’s dataset is more trustworthy, even if both report 99% quality.

📉 Why It Matters: Data Quality Affects Strategy Validity

Even a difference of a few milliseconds in tick granularity can alter trade triggers, trailing stop behaviors, or entry/exit filters.

✅ How to Choose the Best Tick Data for Backtesting

1. Use Your Live Broker’s Data (If Available)

• Always test with the same broker you're planning to trade with.

• This ensures your backtest reflects actual spreads, execution delays, and data quirks.

2. Compare Tick Density

• Visually inspect candles in the Strategy Tester with “Every Tick” mode.

• Look for differences in tick frequency, spread movement, and volume.

3. Use High-Quality Historical Data Sources (if needed)

• Tools like Tickstory, Birt’s CSV2FXT, or commercial data feeds (like Dukascopy) offer richer tick streams.

4. Check Trade Distribution

• If one broker’s data triggers more trades than another, investigate why.

• You may be backtesting against a smoothed-out market environment.

🧠 Pro Tips for Realistic Backtesting in MT5

• ✅ Always enable: “Every tick based on real ticks”

• 🔍 Watch visual backtests with tick-by-tick mode to identify slippage, spread changes, and spikes.

• 📊 Use Monte Carlo simulation to test robustness under different market "noises."

• 🧮 Compare results across brokers with the same EA to assess sensitivity to data variation.

• 🔧 Avoid over-optimization if you’re testing on filtered or low-density tick streams.

📌 Summary: What Makes “Good” Tick Data?

🔚 Final Thoughts

Don’t let a "99% modeling quality" label give you false confidence. The depth, granularity, and realism of tick data are far more important than the label itself—especially if you're developing high-precision strategies like scalping or trading news events.

If your EA relies on precise price behavior, treat tick data as your foundation. A strong strategy on bad data is an illusion. But a solid backtest on reliable data? That’s how real traders build confidence.