Modelling quality is the Strategy Tester's own estimate of how faithfully it recreated real price movement while it ran. It sits at the top of every MT4 report as a percentage, and most traders scroll straight past it. That is a mistake. The figure decides whether the rest of the report is worth reading at all — a beautiful backtest built on poor data is just a well-drawn work of fiction. Here is what the number means and when to trust it.
What it is
MT4 grades each backtest from 0% to 90% and prints the result as its "modelling quality". The percentage reflects how much real tick detail the tester had to work with. At the bottom of the range it uses only the open, high, low and close of each bar. Higher up, it pulls in smaller timeframes to fill in the shape of each candle. The ceiling is 90% — MT4 can never claim a perfect model — and a red bar at the start of the report warns you when the data was thin.
Why the number caps at 90%
MT4 does not store real ticks. To simulate price inside a bar, it interpolates: it draws a smooth path between a handful of control points taken from lower timeframes. Even at 90%, most of the ticks your EA "sees" were invented by that guess, not recorded from the market.
This matters enormously for some strategies and barely at all for others. A slow trend EA that checks price once a bar is largely unaffected. A scalper or a grid EA that reacts to every small move is being fed a tidy, invented move that never spiked, never gapped and never hit the stop the real market would have hit. The interpolation flatters them.
How poor data hides losses
Picture a grid EA that opens a fresh trade every 20 pips against the move and closes the basket at a small profit. In interpolated data, price glides smoothly from one control point to the next, so the basket almost always closes green. In real ticks, price often stabs 15 pips further than the smoothed path showed, triggers one more level, and turns a tidy win into a margin call.
The backtest never sees that stab. So the equity curve climbs in a clean line, the drawdown looks modest, and the real risk — the one that empties the account — is simply missing from the report.
What MT5 does differently
MT5 replaces the percentage with a "history quality" bar and, more importantly, offers a real-tick model. If your broker supplies genuine tick data, MT5 can replay the actual recorded ticks rather than guessing between them. That is a real step up, but only if the underlying data is complete — a real-tick run on patchy history is no better than MT4's interpolation. Check the bar, and check the data source behind it, before you trust the run. It is common to see a strategy that looked superb on 90% MT4 data behave quite differently under MT5 real ticks, usually worse. If the two disagree, believe the ticks.
What a trustworthy backtest needs
- Real ticks — recorded market data, not interpolated bars. On MT4 that means 90% and quality data imported first; on MT5, the real-tick model.
- A real variable spread — spread that moves as it does live, not a fixed one-pip fantasy.
- Commission and swap — the costs you actually pay, applied to every trade.
Miss any of these and the report describes a market that does not exist. Get all three and you have something worth forward testing — still not proof, but no longer fiction.
The Karnek note
Modelling quality decides whether a backtest is worth believing; live results skip the question, because there is nothing left to model. Karnek reads your EA's real trades from the terminal — real ticks, real spread, real slippage — so the numbers are the market's, not a simulation's. It is read-only and can never place a trade, so it reports the truth without touching it.
Written and reviewed by the Karnek Research team. Last updated August 2026.
Educational content only - not financial advice. Past performance does not predict future results. Trading carries significant risk.