Tick Data vs Spread in a Backtest: Bar Order Decided 20 of 4,282 Scalping Trades, the Spread Hit Every One

If you have spent any time on forex forums you have seen this: someone posts a backtest, and the first reply is “what was your modelling quality?” The idea is that a test on 1-minute or 5-minute bars cannot know whether price hit the stop or the target first, so only a tick-data test can be trusted.

That is true in principle. The question is how often it matters. I took the same six rules as in my backtest vs demo account test and counted.

What exactly did I compare?

• Data: EURUSD M5, June 3 to August 26, 2026, 16,926 bars.
• Rules: EMA 9/21 cross, RSI 14 leaving 30/70, Bollinger 20/2 fade, MACD 12/26/9 cross, Supertrend 10/3 flip, Stochastic 14/3/3 cross beyond 20/80.
• Stop and target: equal size, 3, 5, 8, 10 and 15 pips. That is 30 setups. Exit after 24 bars if neither is hit.
• Best bar order: if one bar touches both levels, the target counts. This is what a careless bar backtest does.
• Worst bar order: the stop counts. This is the most pessimistic answer tick data could give.
• Spread: 1 pip, 2 pips from 21:00 UTC, plus 0.3 pips slippage on stops, entry at the next bar’s open. The spread figures come from my spread and swap measurement.

Tick data can only move a trade between the best and the worst bar order. So the gap between those two columns is the most tick data could ever change on these trades.

How often does one bar hit both the stop and the target?

A 5-minute EURUSD bar has to cover the whole distance from stop to target to create the problem: 6 pips for a 3-pip setup, 30 pips for a 15-pip one. The average M5 bar in this sample is a few pips from high to low, so a bar that big is rare. When it does happen, it is at two times of day. Of the 20 double touches at 3 pips, 10 came between 20:55 and 21:30 UTC, the daily rollover when spreads widen and prices jump, and 7 came between 12:25 and 13:40 UTC, the US data hour. My news release test shows how much bigger the 12:30 UTC candle is.

Does the bar order change the result?

Pips per trade, no spread. At 15 pips RSI moves by 0.12 because of a single trade: one double touch flips from +15 to -15, 30 pips spread over 261 trades.

The biggest gap at 3 to 10 pips was 0.07 pips a trade. No setup changed sign. Win rates moved by at most one percentage point: RSI at 3 pips won 53.2% with the best order and 52.2% with the worst.

And what does the spread do to the same setups?

Pips per trade. “No spread” uses the best bar order; “spread” adds 1 pip (2 late), 0.3 pips stop slippage and next-bar entry.

Here every setup moved by about a pip, and every positive one turned negative. Put the two tables side by side and the ranking of worries is clear:

So when does tick data matter?

• When your stop and target are smaller than a normal bar. I did not test below 3 pips. At 1 to 2 pips the double-touch share will be much higher, and so will the spread’s share; that kind of scalping loses on the spread before bar order comes into it.
• When the EA trades inside the bar: trailing stops, partial closes, grid steps. My grid EA test is the kind of logic where every tick counts.
• Around the 21:00 UTC rollover and the 12:30 UTC data hour, which is where 17 of the 20 double touches in this test happened.

For a rule that enters on a bar close and uses a stop and target of 5 pips or more, the spread setting in your tester matters more than ten times as much as the modelling quality. Set the spread to what your broker really charges, not “current”, which on a weekend can be anything. That step is in my guide to backtesting an EA on MT5, and the stop-size side is in the stop loss distance test.

Frequently asked questions

Do I need tick data to backtest in MT4?

Not for bar-close rules with stops of 5 pips or more. On 16,926 EURUSD M5 bars, bar order decided 12 of 3,772 trades at 5 pips and changed results by at most 0.07 pips a trade.

What does 99% modelling quality mean?

That the tester used real ticks instead of generating price paths inside each bar. It fixes the order of events within a bar. It does not add the spread, which in this test cost 0.8 to 1.4 pips a trade.

Why is my backtest better than my live results?

In this test, mainly the spread. Across 30 setups the spread and slippage turned all 8 profitable ones into losers, while the bar-order question changed none of them.

What spread should I use in the Strategy Tester?

Your broker’s normal spread for the hours you trade, with extra for the late hours. I used 1 pip on EURUSD and 2 pips from 21:00 UTC.

Is “open prices only” good enough for testing?

For rules that only act at a new bar and use stops wider than a normal bar, the bar-order error in this test was under 0.1% of trades, so the cheaper mode is usually fine for a first pass.