Tick Audit - check real tick history and backtest quality month by month - script pour MetaTrader 5

"Why is my backtest showing 0% history quality?" is one of the most repeated questions on this forum, and the answers are always the same words: it depends on your broker, do not test further back than your tick data goes. Both true. Neither of them tells you where your tick data actually goes back to.

That is all this script does. It asks the terminal, month by month, and prints the answer before you configure the tester rather than after.

The problem it is for

The tester will happily run over a period that has no real ticks. It does not stop, it does not warn you, it draws a curve. Behind that curve, the ticks were generated from M1 bars.

Generated ticks move in a straight line inside each bar. Nothing spikes, nothing gaps, the spread never widens. The strategies that survive on that and die on real ticks are the well known ones: grids, martingales, anything scalping a few points, anything with a stop close to the entry.

The dangerous month is not the empty one. It is the month that has bars but no ticks, because that is the one that still produces a result.

What it prints

One line per month, with a verdict: real ticks, partial - holes in the ticks, GENERATED TICKS - bars only, or nothing at all. The screenshot shows a real run on gold, where every month from September 2024 to December 2025 has bars and zero ticks, and real ticks only begin in January 2026.

The last two lines are what most people came for: the month where unbroken real ticks begin, and how many months in the range would force the tester to invent ticks.

Inputs

• Months back - how far to audit.
• Check every day - the thorough pass. Off by default because it is slow.
• Sample days - days sampled per month when not doing a full scan.
• Wait seconds - how long to let the terminal download tick history before starting.
• Write CSV - also save the month table to MQL5\Files.

What it does not claim

Sampling tells you whether ticks exist in a month. It does not prove that every minute of that month is complete. For that, turn on the full scan, which checks every day it finds data for. It is slower, and worth doing once before a serious test.

A day is counted only when it carries bars or ticks, so the script never assumes a five day week. That is what lets it read crypto correctly, where Saturday and Sunday trade normally, and instruments with their own holiday calendars. Months where the terminal holds nothing at all are printed as such rather than left out, because a gap in a table reads as an oversight.

If the symbol is not in Market Watch the script adds it itself. Days are server time. It places no orders and changes nothing.

How to run it

1. Copy TickAudit.mq5 into MQL5\Scripts inside your terminal's data folder. In MetaTrader, File then Open Data Folder takes you there.
2. Open it in MetaEditor and press F7. It should report 0 errors and 0 warnings.
3. Back in the terminal press Ctrl+N for the Navigator, open Scripts, and drag TickAudit onto the chart of the symbol you want to audit. The chart's timeframe does not matter.
4. The inputs dialog appears. The defaults audit twenty-four months by sampling six days from each, which takes a minute or two. Press OK.
5. Read the table in Toolbox, Experts tab.

Run it on the symbol you actually intend to test rather than a similar one. Tick history is stored per symbol, so gold having ticks tells you nothing about EURUSD, and XAUUSD tells you nothing about XAUUSDm on the same account.

On a symbol with a lot of tick history the first run can take several minutes. Asking for a month of ticks makes the terminal download that month before it answers, and a request already in flight cannot be interrupted, so the script says this before it starts. A month it marks as bars-only on that first pass may simply not have finished downloading. On a heavy symbol run it a second time: if the verdict changes, the terminal fetched the ticks in between; if it stays the same, the broker does not have them. Later runs are quick because the cache is warm.

How to use the result

Run it first. Take the date where unbroken real ticks begin, set your test start on or after it, and select "Every tick based on real ticks". If that leaves you too little history to test on, that is a real finding too - it means the confident number you were about to produce was never available to you.