Tick Data vs Minute Data: Scaling for Accuracy
The Logic of the Gap: What is Tick Data?
Tick data represents every single price change (bid/ask) that occurs in the market. In a single minute, there could be hundreds of ticks as liquidity providers battle for price superiority. Conversely, Minute Data (M1) only records the Open, High, Low, and Close (OHLC) of each 60-second period, completely ignoring the 'Chaos' that happens inside the candle.
For long-term trend-following EAs, M1 data is often sufficient. However, for scalpers and high-frequency algorithms that target 2-5 pips, relying on M1 data is the most common cause of 'Fake' backtest results. Without tick-accurate data, your EA is essentially guessing how the price moved from Low to High inside the candle.
The MT4 Trap: How Fake Ticks Are Generated
When you run a backtest in MetaTrader 4 using standard M1 data, the platform uses a 'Random Fractal' algorithm to generate fake ticks within the candle. It assumes that if the OHLC is known, it can interpolate the movement. This is fundamentally flawed.
Because the MT4 tester doesn't know if the High happened before or after the Low within a single M1 candle, it will often 'Fill' your take-profit or stop-loss unfairly. This results in the infamous '90% Modeling Quality'—which is actually a failing grade in professional algorithmic trading. You want 99.9% modeling quality, which is only possible with real, historical tick data.
{ Tick Data Checklist }
Primary: Access to historical tick data (Dukascopy/LMAX)
Precision: Achieve 99.9% modeling quality
Variable: Must test with dynamic (floating) spreads
Slippage: Account for 100-200ms delay in execution
Verification: Compare backtest results with live 'Forward' tests
Tooling: Use Tick Data Suite (TDS) for MT4
The Modeling Quality: Achieving 99.9% Precision
Professional quantitative traders use third-party tools (like Tick Data Suite or QuantConnect) to inject actual historical ticks from providers like Dukascopy or TrueFX into their testing environment.
Real tick data includes Variable Spreads. During major news events or the Asian session rollover, the spread can widen from 0.5 pips to 10 pips. Standard MT4 testing uses a 'Fixed' spread, making your EA look profitable in conditions where it would actually be destroyed by the widening spread and slippage.
Why it Matters: The Scalper's Nightmare
If your EA's average profit per trade is 3 pips, a 1-pip error in your backtest data isn't just a minor mistake—it is a 33% error in your total expectation. This is why so many retail EAs look like 'straight lines' in the backtester but fail immediately when moved to a live account. They were developed on 'Perfect' but fake data.