What Is MT5 History Quality? Real Ticks and Its Limits

Let me start with the conclusion. MT5's history quality is a gauge of how closely the tested data matches actual ticks. It is not a number that tells you whether the EA will keep winning in the future. A 99% means the data is well covered, not that the strategy is good. Conversely, at 5% you should assume that most of the equity curve in the report was drawn on "plausible-looking ticks" that MT5 synthesized from bar data.

As a distributor of EAs, we read these reports almost every day. This article lays out what the quality figure shows and where the number stops being informative.

What Does History Quality Measure?

It Can't Be Separated from the Tester's "Model"

MT5's Strategy Tester offers several models for reproducing price movement:

• Every tick based on real ticks
• Every tick (generated from 1-minute bars)
• 1 minute OHLC
• Open prices only

History quality is most meaningful when you choose the first one, the real tick model. This model tests against the actual ticks stored on the broker's server. However, in periods where no real ticks exist, MT5 fills the gap by generating ticks from 1-minute bars. As we understand it, history quality is a rough measure of "the share of the test period that could be reproduced with real ticks."

99% vs 5%

Low values like 5% typically appear in long-period tests where the broker has not kept old real ticks. Older years are generated from 1-minute bars, and only the most recent stretch is real ticks. Because the equity curve looks like one continuous line, you can't see where the boundary lies.

What Changes with Generated Ticks?

Ticks built from 1-minute bars fill in the "order" of the high and low within the minute, and the "movement in between," by estimation. As a result, the following kinds of EAs tend to produce skewed results:

• Scalping-type EAs with take-profit and stop-loss measured in a few pips
• Setups where both a limit order and a stop order could be filled within the same minute
• EAs with filters that avoid moments when the spread widens

By contrast, trend-following EAs that hold positions for tens of hours on H1 or H4 tend to be relatively less affected by price movement inside a single minute. For example, our MEGAMAX DONCHIAN (USDJPY H1) has an average holding time of 32:58:21, and BITCOIN COMET (BTCUSD H4) has an average holding time of 63:51:39. Even so, we can't say there is "no effect." That is why we put more weight on switching models and comparing results than on relying on the quality figure.

How to Check for Yourself

When choosing an EA, or re-testing a distributed EA on your own machine, go through the following in order. Basic MT5 operation is covered in MT5 setup.

1. Run the test with the model set to "Every tick based on real ticks"
2. Check the history quality in the report (also see whether the value changes when you shorten the test period)
3. Run the same period on "1 minute OHLC" as well, and compare PF and maximum drawdown
   • If they differ greatly, the EA depends heavily on price movement within a single minute
4. Split the test period into a first half and a second half
   • Check whether performance falls apart in the second half alone, where real ticks exist
5. Check the spread setting
   • The real tick model uses the recorded spread. Don't confuse it with results from a fixed spread
6. Check the journal (operation log) for data-shortage warnings

The comparison in step 3 is especially effective. Even with a report showing 99% quality, we believe an EA whose results reverse under OHLC is prone to breaking down when fills drift in live trading.

The Limits of History Quality

From here on, we cover the points that the number alone can't settle, including the inconvenient ones.

• Even 99% is only "that broker's ticks." Different brokers have different spread behavior and different tick delivery frequency. A 99% report on Broker A doesn't guarantee results on an account at Broker B. For differences between brokers, see also our FX broker comparison.
• Slippage and order rejections are not reproduced. Even with real ticks, the tester assumes "the order filled at that price." Slippage around economic releases and the like is not included.
• High quality means little if the period is short. A 99% test covering only the most recent year is not necessarily more trustworthy than a low-quality long-term test, because it contains only a single market regime.
• Overfitting happens regardless of quality. No matter how accurate the data is, results will look good if the parameters are fitted to that period.
• The "true performance" in low-quality segments is unknown. Whether results on generated ticks were better or worse than reality is something there are only limited ways to verify afterward. We don't know either.

Some of our EAs have long-period tests published. For example, GOLD NEURON's period is 2004.01.01 - 2026.09.01. How much of the older data is actual tick data depends on how far back the broker stores it, so please don't read a long period as evidence of high quality. You can see each EA's figures side by side in the measured ranking.

Summary

• History quality is a gauge of how much could be reproduced with real ticks, not an indicator of an EA's merit
• Low values usually mean a long-period test where old real ticks are missing and have been filled with ticks generated from 1-minute bars
• The shorter-term the EA, the more it is affected by generated ticks
• It is more practical to check whether results differ greatly between real ticks and OHLC than to rely on the quality number
• Even at 99%, results at another broker, slippage, and future market conditions are not guaranteed

We will continue covering how to read backtests in EA & MT5 know-how.