Tick data records every price change. Why real ticks make a backtest realistic, how modelled ticks flatter results, and where to get them.

What is Tick Data?

Tick data is a record of every individual price change on an instrument: each new bid and ask, in the order it arrived, with a timestamp. A single one-minute candle can hide dozens or hundreds of ticks. For an Expert Advisor, ticks are the heartbeat. The EA wakes on each one, checks its rules, and decides whether to act. Backtest on realistic ticks and you test the EA on something close to real trading; backtest on invented ones and you test a fantasy.

What it is

A tick is one update to the price. Every time the bid or ask moves, even by a fraction of a pip, that is a tick. Tick data is the full stream of those updates: thousands per hour on a liquid pair during active sessions, far fewer overnight. It is not the same as candle data. A one-minute candle records only four prices: open, high, low and close.

The path the price took between those four points is exactly what candles throw away and tick data keeps. Which came first, the high or the low? How many times did price stab down before it closed up? That path matters, because it decides whether your stop or your target was hit first, and at what price.

How it works

When MetaTrader's Strategy Tester runs a backtest, it feeds the EA a sequence of ticks and lets it react. The question is where those ticks come from. On MT4's older modes there are no real ticks at all: the tester has candles and invents a handful of fake ticks inside each bar to approximate the movement. These are "modelled" ticks, summed up as a modelling quality percentage, with 90% the old ceiling. MT5 and tools such as Tick Data Suite instead replay genuine recorded ticks, so the price walks the path it truly walked.

Real ticks come from a few places: MT5's own history from brokers that store it, third-party archives such as Dukascopy, and importers that load recorded ticks into MT4. The difference from modelled data is not cosmetic. Modelled ticks smooth over the exact moments an EA lives or dies, such as the spike that clipped your stop or the spread widening on news.

A worked example

Take a scalping EA with a 5-pip stop and a 5-pip target. Backtested on MT4 with 90% modelled ticks it shows 68% winners and a tidy equity curve, because inside each modelled bar the price glides politely and tends to reach the target before the fabricated stop. Run the same EA over the same period on real tick data and the picture changes. Real ticks show the price stabbing 6 pips against the position before recovering, so trades that "won" on modelled data are stopped out first. The win rate falls to 51%, and what looked like a £3,000 month becomes a £200 loss. Nothing about the EA changed, only the honesty of the ticks.

Why it matters

Short-stop strategies (scalpers, grid systems, anything trading the spread) are the most sensitive, because a few pips of real intrabar movement decide every trade. The tighter the stop, the more the tick source matters. A backtest built on modelled ticks can turn a losing EA into a winning one on paper, which is precisely the result a vendor selling that EA is happy to show you.

Real tick data also carries real spread and, ideally, real slippage, so the costs in the test resemble the costs you will actually pay. Modelled runs often assume a fixed, generous spread that live trading will never give you.

Common misconceptions

90% modelling quality means 90% accurate. It does not. It measures how much candle data was available to model from, not how close the fake ticks are to reality. A 90% run can still be badly wrong.
MT5 fixes this automatically. MT5 is better, but its "every tick based on real ticks" mode is the one that uses genuine recorded ticks, and only if your broker or data source supplies them.
Tick data is only for scalpers. Any EA with a stop close to entry, or one that trades around news, is exposed. Wide-target trend systems are more forgiving, not immune.
One provider's ticks are as good as another's. Ticks are broker-specific. The feed you test on should match the broker you will trade, or the test drifts from your reality.

The Karnek note

Karnek does not backtest; it reports what your EA does live, tick by tick, on the real account. That makes it the honest counterweight to any backtest: whatever the modelled ticks promised, the live figures are the ones Karnek shows. It reads the terminal read-only and can never place, close or modify a trade.

See it in Karnek: monitor grid and martingale EAs, with drawdown alerts before they run away.