Tick Data vs 1-Minute Data: Same Candle, Different Trade

Here is a small thought experiment.

Suppose a one-minute candle looks like this:

• Open: 100
• High: 110
• Low: 90
• Close: 105

Your strategy buys at 100.

It has:

• take profit at 108,
• stop loss at 95.

Did the trade win or lose?

From the candle alone, you cannot know.

If price travelled:

100 → 110 → 90 → 105

the take profit could have been reached first.

If it travelled:

100 → 90 → 110 → 105

the stop could have been reached first.

The OHLC values are identical.

The trade outcome is not.

A candle records extremes, not the journey

A one-minute bar is excellent at answering:

• Where did the minute begin?
• What was its highest price?
• What was its lowest price?
• Where did it finish?

It does not normally preserve the exact sequence of every price update.

This distinction becomes important whenever the order of intrabar events matters.

That includes more strategies than many traders initially expect.

When M1 can be perfectly useful

Minute data is not “bad data”.

Far from it.

For early-stage research it can be extremely efficient.

Suppose a strategy:

• calculates indicators only on completed bars,
• opens positions only at the next bar,
• uses wide risk limits,
• never reacts to intrabar movements.

Running years of real tick history may add huge processing cost without materially changing the logic being tested.

In that situation, M1 or another bar-based representation can be an excellent first filter.

You can reject weak ideas quickly before spending time on higher-resolution validation.

When M1 begins to lose important information

The problem appears when strategy behaviour depends on events occurring inside the minute.

Examples:

Tight stops and targets

If both were crossed during the same bar, OHLC alone may not tell you which happened first.

Trailing stops

A trailing stop depends on the path.

Price reaching a new high before falling is not equivalent to falling first and recovering later.

Pending orders

A stop or limit order can be activated by a short-lived intrabar movement that disappears completely into the finished candle.

Spread-sensitive systems

An M1 Bid candle does not describe every Ask movement that occurred during the minute.

Intrabar indicators

Some algorithms recalculate indicators after every market update rather than waiting for a candle to close.

For them, the path itself becomes part of the input.

Generated ticks are an estimate of the missing path

MetaTrader 5 offers an “Every tick” testing mode that can generate intermediate ticks from one-minute bar data.

This is useful.

It allows much richer event-driven testing than simply processing four OHLC points.

But generated ticks are still generated.

MetaTrader creates a sequence based on known M1 control points and modelling rules.

The historical tick path itself is not recovered.

By contrast, “Every tick based on real ticks” uses recorded tick history where it is available.

That difference is crucial:

generated ticks answer “what path can we simulate from this candle?”

real ticks answer “what market updates were actually recorded?”

More resolution does not always mean more truth

Even real tick datasets come from somewhere.

For spot FX, there is no single consolidated tape showing every quote from every market participant.

Different brokers and liquidity venues can legitimately record slightly different streams.

So there are two separate questions:

1. How much resolution does the dataset contain?
2. How representative is that dataset of the trading environment being simulated?

A beautifully complete external dataset can still differ from your broker's live feed.

That does not make the dataset useless.

It means its role has to be understood correctly.

A better backtesting workflow

Instead of arguing about one “correct” testing mode, use different levels of fidelity for different jobs.

Stage 1 — Fast research

Use bars or simplified tick generation to reject obviously weak ideas.

Stage 2 — Detailed validation

Retest surviving strategies using real historical ticks.

Stage 3 — Environment matching

Check spread behaviour, broker symbol properties, sessions, time mapping and costs.

Stage 4 — Out-of-sample testing

Test data the strategy was not optimized on.

Stage 5 — Demo or controlled forward observation

Compare simulated behaviour with a live market feed.

This workflow preserves speed without pretending that speed and realism are the same thing.

The most dangerous backtest is not necessarily the inaccurate one

It is the one whose assumptions you do not understand.

An M1 test can be completely appropriate if its limitations match the strategy.

A 100% tick-quality test can still mislead you if the time zone, symbol specification, spread model or execution assumptions are wrong.

Data resolution is therefore not a score to maximize blindly.

It is a tool to match to the strategy.