Every MT5 Strategy Tester report opens its results with one line that most people read past: History Quality. It is a percentage, followed by the words “real ticks”. It tells you how much of your test ran on prices your broker actually recorded, and how much ran on prices the tester made up.
Here is that line from two of our own reports. Both are the same EA, Tokyo Fix on AUDJPY M5, with the same settings, tested over two different windows. Our test terminal runs in Japanese, so the labels are Japanese; the English report says History Quality, Bars and Ticks.
2019.01.01 – 2026.09.05:
ヒストリー品質:
8% リアルティック
バー:
573233
ティック:
3255959792010.01.01 – 2018.12.31:
ヒストリー品質:
0% リアルティック
バー:
666826
ティック:
238486872Three hundred million ticks, and 8% of them real. The older test has no real ticks at all. This article explains what that number measures, why ours is so low, and how much it actually changed our results.
It belongs to our guide to how to backtest an EA in MetaTrader 5, step by step.
What History Quality Measures
The Strategy Tester needs a price for every moment of the test. It has two places to get one:
- Real ticks — the bid and ask your broker’s server recorded, with the spread that was actually quoted.
- Generated ticks — prices the tester builds from one-minute bars. Each generated path respects the bar’s open, high, low and close, but the order and timing of the moves inside the minute are a model.
History Quality is the share of the test that came from the first source. 100% means every price was recorded. 0% means every price was built from bars.
It is not a score of the strategy, and it is not a score of the backtest. It says nothing about slippage, requotes, or whether your logic reads the current bar. Those are covered in why 99% modeling quality can still mislead. History Quality only answers: were these prices real?
Why Ours Is 8%, and Why Yours Might Be Too
The tester can only replay ticks the broker’s server has kept. Our broker’s tick history begins on 2 January 2026 (and for USDJPY, 10 February 2026). Anything before that is generated, whatever mode you choose.
The arithmetic matches the report. The 2019–2026 window runs about 92 months, and roughly 8 of them have recorded ticks. That is the 8%. The 2010–2018 window has none, so it reads 0%.
This surprises people, because they selected “Every tick based on real ticks” and expected real ticks. The mode is a request, not a guarantee. Where the history does not exist, the tester fills the gap with generated ticks and runs anyway. Our test runner says this in its own header:
Model codes: 0 every tick (generated), 1 M1 OHLC, 2 open prices, 3 math, 4 every tick based on
real ticks. The report's History Quality still says what was actually delivered.So read the percentage, not the mode you picked. Two tests run in the same mode can sit on very different data, and only this line tells you which.
How Much Did It Change Our Results?
Less than you might fear, for the kind of EAs we run. When we ran a 280-trade M15 test on generated ticks and again on partly real ones, the difference was 2 trades and 0.00 in profit factor.
The Tokyo Fix EA above shows the same thing from another angle. It enters six seconds after the 09:55 Tokyo fix and holds for 25 minutes. When we tested adding that 6-second delay, it had no detectable effect on generated ticks in either direction. A strategy that holds for minutes does not live or die on the shape of a single minute.
That is a statement about our EAs, not about all EAs. History Quality matters a great deal when:
- The stop or target is a few pips. Generated paths inside a minute decide which one was hit first.
- The edge is close to the spread. When we planned our first screening runs, we moved them to real ticks for exactly this reason: on a tight-stop system, the real spread is most of the edge.
- The EA reacts to individual ticks rather than to closed bars.
If your EA is any of those, a test at 0% real ticks tells you very little, and you should say so when you report it.
What We Do With It
We made one decision early: History Quality is disclosed, never used as a gate. A test at 0% is admissible. It is simply labelled.
The reasoning is practical. If only real-tick tests counted, we could only test the last ten months, and ten months is far too short to see a strategy fail. Our 2010–2018 Tokyo Fix test, 210 trades at 0% real ticks, is the out-of-sample window that the 2019–2026 result has to survive. Throwing it away to keep the percentage high would make the evidence weaker, not stronger.
So every backtest we publish carries its tick source next to its numbers. On the Tokyo Fix AUDJPY page, and in the JSON behind it, each run says Mixed with its percentage, or EveryTick for fully generated history. A profit factor without that label is a number you cannot interpret.
How to Read It in Your Own Reports
- Find the line before anything else. It is the first line of the results.
- Compare it with your date range. If the percentage is far below what your mode suggests, your broker’s history is shorter than your test.
- Ask whether your EA is sensitive to it. Tight stops, small edges and tick-level logic: yes. Minutes-long holds on closed bars: much less.
- Report it with the result. “PF 1.37 on 210 trades, generated ticks” is a claim someone can check. “PF 1.37” is not.
The number will not tell you whether a strategy works. It tells you what the test was made of, and that is the first thing you need to know before you believe anything else in the report. The rest of the report is explained field by field in MT5 Strategy Tester Report: Every Field Explained.
Comentários