Tick Data

Tick data quality is one of the most underdiscussed factors in EA evaluation. Two strategies running identical logic can produce wildly different backtest results depending on the tick-data source's accuracy and density.

What tick data records:

• Every bid price change with timestamp (some sources also record ask price changes) • Sometimes: volume per tick (number of trades or share of dealer interest) • Sometimes: bid-ask spread at each tick • Per major pair: 5,000-20,000 ticks per trading day in normal conditions, higher during news events

Tick data sources for EA backtesting:

• **MetaTrader 5 built-in**: MT5 ships with broker-provided tick data; quality depends on the broker. IC Markets Raw provides high-quality tick data; some Standard-account brokers ship low-density approximations that don't accurately represent micro-structure • **Dukascopy historical data**: free, high-quality, multi-year coverage; used as the editorial benchmark for tick-data-quality reference • **Tickstory and Tick Data Suite**: commercial historical-data services with millisecond timestamps and explicit spread data; the industry standard for backtest quality assurance • **Direct from prime brokers**: institutional traders sometimes acquire tick data directly; not relevant for retail EA evaluation

Why tick data matters for EA backtests:

• **Scalping strategies are tick-sensitive**: M1 scalping strategies producing entries on tick-level price action will backtest very differently on real tick data vs on M1-bar approximation. The bar approximation assumes a smooth price path between OHLC points; real tick paths can include rapid reversals that hit stops the bar didn't show • **Slippage modelling requires tick density**: realistic slippage assumes the strategy can be filled at the tick prices that existed at submission time; thin tick data forces backtest models to assume best-case fills that don't reflect reality • **Spread modelling requires ask data**: many tick datasets record only bid prices; backtests then have to estimate ask prices, introducing spread-modelling error that compounds for high-frequency strategies • **99% modelling quality vs 90%**: MT5 Strategy Tester reports modelling quality as a percentage; 99% quality (tick data) produces backtest results that approximate live execution; 90% quality (bar data with approximated ticks) produces noticeable divergence

For EA evaluation, vendors who backtest using broker-provided tick data on Tier-1 ECN brokers (IC Markets Raw, Pepperstone Razor) produce backtests that approximate live execution. Vendors who backtest using bar data or low-quality tick approximations produce results that don't generalise to live deployment. The MT5 Strategy Tester's modelling-quality percentage is the simplest single-number indicator of tick-data quality used in a backtest — anything below 99% materially diverges from live execution for scalping strategies.