The truth about tick data in MT5 backtests: neither ‘Every Tick’ nor ‘Real Ticks’ are perfect
The truth about tick data in MT5 backtests: neither ‘Every Tick’ nor ‘Real Ticks’ are perfect
For years, across forums, marketplaces, and private groups, one idea is repeated as if it were an absolute truth:
"If you use Every Tick data or high-quality historical data, your backtest is reliable."
No.
It is not.
And this belief—which has become almost a religion—has destroyed more accounts than any poor strategy.
Today I want to dismantle it from its technical foundations and show you an uncomfortable truth: The backtest doesn't lie to you. You lie to yourself when you interpret it as reality.
1. The Technical Myth: "Every Tick" = Absolute Truth (The Idealized Simulation)
Many traders believe that by activating "Every Tick Based on Real Ticks" in MT5, they have a perfect market simulation.
But the Strategy Tester doesn't reproduce the market; it reproduces an idealized, mathematically logical version of the market. And those are not the same thing.
✔ What MT5 actually does:
-
It reconstructs a sequence of ticks based on the raw history.
-
It distributes prices, highs/lows, and volume according to an internal algorithm.
-
It smooths out gaps and micro-gaps.
-
It uses a local server without real market latency, slippage, or friction.
In other words: It does not model market chaos. It models an "orderly" market according to the tester's rules. And you optimize your EA for that perfect world... but you're going to trade in one where nothing is perfect.
2. Tester Ticks vs. Market Ticks: Barely Cousins
A real tick is the result of the order flow reaching the broker's server. A tester tick is the result of an algorithm that "fills in" the past to make it simulatable.
Key Differences:
-
A. Real Ticks are non-linear; Tester Ticks are. In the real market, there are tick clusters, 300ms pauses, or explosions of 50 ticks in 50ms. In the tester, everything flows with perfect cadence.
-
B. Real Ticks contain "structural noise." False spikes, micro-gaps, liquidity inconsistencies. In the tester, that noise magically disappears.
3. Silent Latency: The Invisible Algorithm Killer (And Why MT5's "Delay" Setting Fails)
The tester's simulated execution is immediate. While MT5 offers the "Delays" field to simulate latency (like the 1 ms in your settings), this adjustment is, ironically, another way to create an illusion of certainty.
By setting a fixed delay, you are assuming a constant, perfectly predictable latency in every execution.
The Live Reality is the opposite: The total execution time is the sum of: VPS Latency + Network Congestion + Broker Queue + Real Slippage.
That sum is not fixed; it is a stochastic variable that changes with network traffic and volume. The real market confronts you with a dynamic enemy that can range from 5 ms to 300 ms in a volatile instant. You are optimizing for idealized latency, not for real latency.
4. Dynamic Spread: The Destroyer of "Perfect" Strategies
Many EAs are optimized assuming a fixed spread. But in reality, even on ECN accounts, the spread "breathes." News events destroy it, and nocturnal liquidity twists it.
What happens if your bot opens trades when the spread is 1 point above the assumed average? Everything changes: the R:R, the exact point of invalidation, and the real probability of hitting your SL or TP. Many bots fail because they were designed for a lower execution cost than the real one.
5. The Silent Theft: Believing in the Illusion
The tester doesn't steal your money. It steals your perception.
A backtest with 99.9% Model Quality seduces you and gives you false emotional certainty. It makes you believe your system is ready for real war.
The EA didn't fail. Your expectation, based on a simulated universe, failed.
The Mindset Shift: From Optimization to Robustness
Stop! We've reached the turning point. If the backtest lies, how do we test?
The goal ceases to be finding the parameters that give the prettiest equity curve, and shifts to finding the design that survives the largest number of adverse conditions and imperfect data.
We stop seeking historical perfection to seek tolerance for chaos.
The Testing Dilemma and the Solution
We acknowledge a fundamental dilemma, especially when buying on the MQL5 Marketplace:
-
Dilemma: The only way to validate an EA before buying or renting is through the backtesting provided by the vendor.
-
Risk: If we blindly trust that idealized curve, the risk when transitioning to live trading is maximum.
-
Solution: You must use the backtest only as an initial filter and move to Forward Testing as the mandatory validation.
Forward Testing is the only way for your EA to face the real latency, dynamic spread, and actual execution of your broker.
-
For Purchased/Rented EAs: Use the rental period to mandatorily test the EA on a LIVE Demo/Simulation account with the same broker you plan to use.
-
For Free/Proprietary EAs: Do not take it to a real account until it has passed a Live Demo Forward Test period (months, not days) that demonstrates survival.
7. Robustness Tactics (How to Survive Real Friction)
Here is the plan for designing systems tolerant to chaos:
✔ 1. Anti-Friction Design: Develop your EA to be tolerant of slippage and dynamic spread. Can your entry logic consistently withstand 2-3 pips of worse execution? If not, it's too fragile.
✔ 2. Multi-Feed Optimization: Never optimize on just one history. Test on data from different brokers (if ethically possible) or at least, on several extreme variable spread ranges within the tester. If it only works on a perfect feed, it doesn't work.
✔ 3. Prioritize Logic over Precision: Strategies that rely on the surgical accuracy of a tick for their invalidation or entry are the first to die live. Look for the structural logic of the market (flows, levels) rather than price exactitude.
✔ 4. Forward Test is the Finish Line: Use your Demo/Simulation account as your final battlefield before risking capital. It is your only path to real validation.
Conclusion
The problem is not MT5. The problem is thinking the tester is a perfect time machine.
The backtest is an approximate simulation of the market. And until you understand that difference and adopt the mindset of Robustness and Forward Testing, every perfect optimization will be a trap, and every idealized curve will be a mirage.
Your real proving ground is not the data history folder; it's the live execution of your broker.
The Holy Grail of automated trading is a myth perpetuated by marketers. It simply does not exist.
