The MT5 backtest report, explained like you are five
Read an MT5 backtest report in this order: how was it tested (modelling and history quality), how much it can hurt (equity drawdown), whether profit is worth the pain (profit factor and recovery factor), and whether there is enough evidence (total trades). Net profit comes last. Below, every number in the report is explained with a lemonade stand — simple enough for a five-year-old, accurate enough for a trader.
The lemonade stand
Imagine your robot runs a lemonade stand for a year. Every cup it sells is a trade. Some cups make money (sweet lemonade), some lose money (sour lemonade). The backtest report is the notebook where the robot wrote down every cup. Reading it is just asking a few simple questions in the right order.
The examples below use made-up numbers to explain the ideas — they are not results of any real EA.
Step 1: How was the test done? (the top block)
Before believing any number, check the rules of the game.
| Field | What it means | Like you are five |
|---|---|---|
| Modelling | How the tester rebuilt prices: real ticks, generated ticks, 1-minute OHLC or open prices only | Did the robot practise with the real lemons, or with drawings of lemons? |
| History Quality | Share of correct versus incorrect one-minute data in the test | How many pages of the old diary were readable |
| Bars / Ticks | How many bars and price updates were simulated | How many customers walked past the stand |
| Symbols, period, dates | What was traded and for how long | Which street and which months |
| Initial deposit, leverage | Starting money and borrowing power | Pocket money at the start, and how much the bank lends |
Look for: "Every tick based on real ticks", a high history quality, a realistic delay, and a period of several months or more. If the modelling is "Open prices only" for an EA that scalps, stop reading — the rest of the report cannot be trusted. Why real ticks matter.
Step 2: How much can it hurt? (drawdown)
Drawdown is how far the money fell from a high point before it recovered. It is the most important risk number in the report. MT5 shows it three ways, for both balance and equity:
- Absolute — how far below the starting money it ever went. Like five: "I started with 10 coins and once had only 7, so absolute drawdown is 3."
- Maximal — the biggest fall from any peak to the next low, in money (the percentage is in brackets).
- Relative — the biggest fall as a percentage of the peak it fell from (the money is in brackets).
Read equity drawdown, not balance drawdown. Balance only changes when trades close; equity includes trades that are still open and losing. A robot that keeps losers open can show a tiny balance drawdown and a frightening equity drawdown. Like five: balance is the coins in your pocket, equity is the coins in your pocket plus the lemonade you still owe people.
Look for: a maximal equity drawdown you could live through without switching the robot off. If it says 45%, picture almost half the account gone.
Step 3: Was the profit worth the pain?
| Field | What it means | Like you are five |
|---|---|---|
| Total net profit | All wins minus all losses | Coins at the end minus coins at the start |
| Gross profit / gross loss | Sum of all winning trades / sum of all losing trades | All the sweet cups added up / all the sour cups added up |
| Profit factor | Gross profit divided by gross loss | For every 1 coin lost, how many coins came back. 1.0 = break-even |
| Expected payoff | Average result per trade | How much one cup earns on an average day |
| Recovery factor | Profit divided by maximal drawdown | How many times the robot earned back its worst fall |
| Sharpe ratio | Return divided by how much returns jump around | Was the money earned calmly or on a roller-coaster? |
Look for: a profit factor clearly above 1 over many trades (a modest 1.3 over hundreds of trades says more than 4.0 over thirty), a recovery factor comfortably above 1, and an expected payoff larger than the costs you will pay per trade. MetaTrader's own help calls a Sharpe ratio of 1.0 or more good and 3.0 or more very good; anything below 0 lost money.
Step 4: Is there enough evidence?
| Field | What it means | Like you are five |
|---|---|---|
| Total trades | Trades that closed with a profit or loss | How many cups were sold |
| Total deals | Every execution, including partial closes | Every time money changed hands |
| Profit trades (% of total) | Win rate | How many cups were sweet |
| Short / long trades (won %) | Sells and buys, and how many of each won | Lemonade vs. orange juice — is the robot good at both? |
| Largest / average profit and loss trade | The single biggest win and loss, and the averages | The best cup, the worst cup, and a normal cup |
| Maximum consecutive losses | Longest losing streak and its money | The most sour cups in a row |
Look for: a few hundred trades at least. Compare the average loss with the average win: a 90% win rate is meaningless if one loss wipes out twenty wins. And picture the longest losing streak happening on your first week — would you keep going?
Step 5: The equity curve and the rest
The graph at the top shows balance (blue) and equity (green). A healthy curve climbs with small dips. Warning shapes are a perfectly straight line (often a grid or martingale hiding open losses), long flat periods with sudden drops, or a curve that only rises because of one or two lucky trades.
The remaining fields are for deeper checks: LR correlation (close to 1 means a smooth, steady curve), LR standard error (how far the curve wanders from a straight line), margin level (the lowest free-margin percentage — very low means the account nearly ran out of room), Z-score (whether wins and losses come in streaks), and AHPR/GHPR (average growth per trade; GHPR is the more honest of the two).
A 60-second checklist
- Modelling = every tick based on real ticks? History quality high? Delay above zero?
- Maximal equity drawdown — could you sit through it?
- Profit factor above 1 over hundreds of trades? Recovery factor above 1?
- Average loss vs. average win — does one loss erase many wins?
- Longest losing streak — survivable at your lot size?
- Curve shape — steady climb, or suspiciously straight?
Want this report for your own EA and settings? The MT5 backtest report service runs your .ex5 and .set file on real ticks and sends one report per timeframe to Telegram. Then check it against the 12 backtest red flags before you go to a demo account.
Frequently asked questions
What is a good profit factor in an MT5 backtest?
Anything clearly above 1.0 over hundreds of trades. Around 1.3 to 2.0 on a large sample is healthy; a very high profit factor on a few dozen trades usually means the settings were fitted to the test period.
Which drawdown should I look at in an MT5 report?
Maximal equity drawdown. Equity includes open losing trades, so it shows the real worst point; balance drawdown can hide losses that were still open.
What does recovery factor mean in MT5?
Net profit divided by maximal drawdown. It shows how many times the EA earned back its worst fall; above 1 means it made more than it lost at its worst.
What is a good Sharpe ratio in the MT5 tester?
MetaTrader help describes 1.0 or more as good and 3.0 or more as very good. Below 0 means the test lost money.
How many trades does a backtest need?
A few hundred closed trades is a sensible minimum before the statistics mean much. Fewer trades can be dominated by luck.
What is the difference between total trades and total deals?
Total trades counts positions that closed with a profit or loss. Total deals counts every execution, including partial closes and entries.