MT5 Strategy Tester modeling options explained: real ticks, test models and historical data quality

What the MT5 Strategy Tester modeling options actually do

The modeling option in the MT5 Strategy Tester decides how much price detail your expert advisor sees during a backtest. It ranges from "Every tick based on real ticks", which replays the broker's recorded tick history, down to "Open prices only", which shows the EA one price per bar. The same EA on the same EURUSD dates can report noticeably different results depending on this choice, the quality of the history behind it and a few other settings in the tester.

This article compares the test models and the data and configuration choices that move your results. For the full step-by-step run and report walkthrough, see how to test an expert advisor in MT5.

The five modeling modes at a glance

You choose the model in the Strategy Tester Settings tab, in the Modeling dropdown. MetaQuotes documents each mode in the MetaTrader 5 testing help.

How each test model builds price

Every tick based on real ticks

This mode downloads the tick history your broker stores on its server and feeds it to the EA in sequence. Each tick carries its own bid and ask, so the spread widens and narrows as it did historically, for example around the 8:30 New York news window on EURUSD or the cash open on an NQ CFD.

Two limits remain. First, it only replays prices; it cannot know whether your order would have been filled at that price in a real queue, so slippage and liquidity are not modeled the way a live futures DOM would show them. Second, if real ticks are missing for some minutes, the tester fills those gaps with generated ticks, which lowers the historical data quality figure covered below.

Every tick

Here the tester generates ticks from M1 bars. It builds a plausible path inside each minute using the bar's open, high, low and close as reference points, and the number of ticks is tied to the bar's tick volume. The path is synthetic: the order in which the high and low were reached inside a minute is estimated, not recorded. If your EA uses a 5 pip stop and a 5 pip target on EURUSD, which one is hit first inside a single minute may differ from reality.

1 minute OHLC

The EA sees only four prices per minute bar. This is much faster and fine for logic that reacts to closed bars or wide stops. It becomes unreliable when stops, targets or trailing logic can both trigger inside the same minute, because the tester has no information about the path between those four points.

Open prices only

The tester produces a single price per bar of the selected timeframe. It is valid only if your EA explicitly works on new bar opens and does not manage positions tick by tick. An EA with intrabar trailing stops tested this way can produce a report that has little to do with how it trades live.

Math calculations

No history is loaded and no trades are simulated against prices. This mode exists for running calculations through the optimizer. It is not a backtest of trading behavior.

MetaQuotes explains the tick generation logic in more detail in its tick generation documentation.

Historical data quality: what the report is telling you

The backtest report includes a history quality percentage along with the number of bars and ticks used. Read it before you read profit factor or drawdown.

• In real ticks mode, a value below 100% means some minutes were not covered by matching real ticks and were generated instead. A few percent can be acceptable for a swing EA; for a scalper, it means part of the test was synthetic exactly where precision matters.
• In generated modes, the figure reflects the completeness of the minute history used to build ticks. Gaps in M1 data mean gaps in the simulated market.
• The tick count is a sanity check. A real ticks test on EURUSD over a year should involve far more ticks than an OHLC test of the same period. If the numbers look too close, the tester likely did not have the tick history you expected.

To inspect what your broker actually provides, open View >> Symbols (Ctrl+U), select the symbol and use the Bars and Ticks tabs to request history for your test dates. Check where the tick history starts and look for missing days. Brokers differ in how far back they store ticks, and data for index or oil CFDs that track ES, NQ or CL futures can be shorter or thinner than for major forex pairs.

One more check: if you compare tests across machines or brokers, the history itself is different. Two brokers' EURUSD feeds will not have identical ticks or spreads, so identical settings can still yield different reports.

Test configuration that changes results besides the model

Delays and execution

The Settings tab includes an execution delay option. Zero latency gives ideal fills at the tested price, which flatters EAs that trade often or react to fast moves. A random or fixed delay is closer to live conditions. Run your final test with a realistic delay, and compare it with the zero latency run: a large gap tells you the strategy depends on speed it may not have live.

Spread and symbol specifications

In generated modes the spread comes from the minute history, so it is smoother than the real tick-by-tick spread. Also confirm the symbol's contract specifications in the Symbols window: contract size, tick size, commission handling and swap. A strategy with small targets is sensitive to costs, which is why a maximum spread filter matters; see MT5 expert advisor risk settings explained for how that safeguard works.

Date range, timeframe and deposit

The chart timeframe in the tester does not change the price resolution in tick modes, but it changes when bar-based logic fires. The date range decides which market regimes are included; a test covering only one trending period says little about a ranging one. Deposit and leverage affect position sizing if your EA uses percentage risk, so keep them consistent when comparing runs.

A practical workflow to pick a model

1. Read the EA documentation to see whether it trades on new bars or manages positions on every tick.
2. Use 1 minute OHLC or Open prices only for fast exploration, only if the logic matches that mode.
3. Download tick history in the Symbols window and confirm it covers your full date range.
4. Rerun the final candidate with Every tick based on real ticks and a non-zero delay.
5. Compare the two reports. Large differences in trades taken, drawdown or average trade point to intrabar sensitivity.
6. Note the history quality and tick count with each saved report so you know what data stood behind it.
7. Forward test on a demo account before trusting any of it; the guide on MT5 demo testing an expert advisor shows how to log differences.

If you are evaluating a ready-made robot, an MT5 expert advisor with documented inputs and set files makes this process easier, because you know which mode its logic is designed for and can reproduce the same configuration.

Conclusion

The modeling mode, the history behind it and the execution settings together define what a backtest report is worth. Real ticks with good history quality and a realistic delay give the most honest replay MT5 offers, but it is still a replay of prices, not of fills. Treat faster modes as screening tools, verify your data before trusting a number, and remember that past simulated results do not predict live performance.

FAQ

Why do I get different results with the same EA and dates on another computer?

The tick and bar history is downloaded per broker and per terminal. If one machine has more complete history, a different broker or a different delay setting, the simulated prices differ and so do the results.

Does visual mode change backtest results?

Visual mode only displays the test on a chart. Results should match a non-visual run with the same settings and data, though it runs slower.

Can I backtest real futures like ES or CL in MT5?

Only if your broker offers those instruments in MT5 with stored history. Many brokers offer CFDs that track ES, NQ or CL instead, with their own prices, spreads and contract sizes, so check the symbol specifications before drawing conclusions.

Is 100% history quality enough to trust a backtest?

It means the data was complete for the chosen mode, not that the strategy is robust. You still need realistic costs and delay, an out-of-sample period and a demo forward test.

If you want an automated strategy whose inputs and test setup are documented from the start, SierraOnShop offers an MT5 expert advisor built with that transparency in mind.