Performance metrics
How to Read an MT5 Strategy Tester Report, Field by Field
How to read the MT5 Strategy Tester report: net profit, profit factor, expected payoff, recovery factor, Sharpe ratio, drawdowns and History Quality explained.
The MetaTrader 5 Strategy Tester report packs dozens of statistics into one page. A few of them carry most of the information, several are easy to misread, and the most important facts about the test sit in the settings rather than the results. This guide goes through the report in the order worth reading it, using the field names and definitions in the MetaQuotes documentation.
Start with the test settings
Before looking at profit, check how the prices were produced. MT5 offers several tick modeling modes:
| Mode | What it does |
|---|---|
| Every tick based on real ticks | Replays tick history collected by the broker; no simulation |
| Every tick | Generates ticks from one-minute bars |
| 1 minute OHLC | Uses only the open, high, low and close of each minute bar |
| Open prices only | Uses bar open prices; fast but coarse |
| Math calculations | No price history at all; for pure calculations |
The difference matters for any strategy with tight stops or short holding times. According to the MetaQuotes documentation on tick generation, when ticks are generated within a minute bar the tester uses the spread fixed for that bar, while real ticks let the spread change inside the minute. Generated ticks therefore smooth over exactly the spread spikes that hurt scalping systems. Where a minute bar has no real tick data, the tester falls back to generating ticks for it.
History Quality is the MT5 field that reports the share of correct versus incorrect one-minute data used in the test. Traders coming from MT4 often look for "modelling quality", the nearest equivalent in the older platform. A low History Quality figure means part of the price history was missing or flawed, so the results deserve less trust.
Also check the execution delay setting. With no delay, orders fill at the requested price without requotes, which flatters any strategy sensitive to slippage.
Profit fields
- Total Net Profit: the financial result of all trades.
- Gross Profit and Gross Loss: the sums of all winning and all losing trades.
- Profit Factor: gross profit divided by gross loss.
- Expected Payoff: the average result per trade, in deposit currency.
These are the headline numbers, and each has a common trap. Net profit depends on lot size and on whether the EA compounds, so a large figure can come from aggressive sizing rather than a strong edge. Expected payoff is in money, so it should be compared with the typical amount risked per trade, not read in isolation. Profit factor needs the trade count next to it. The guide to expectancy and profit factor covers how to judge both.
Risk-adjusted fields
- Recovery Factor: profit divided by maximum drawdown. A value of 4 means the test made four times its worst drawdown.
- Sharpe Ratio: return minus the risk-free rate, divided by the standard deviation of return. MetaQuotes states that the tester assumes a risk-free rate of zero.
- AHPR and GHPR: the arithmetic and geometric mean change in equity per trade.
- LR Correlation and LR Standard Error: how closely the balance curve follows a straight regression line. A correlation near 1 means a steady curve.
- Z-Score: a test of whether wins and losses cluster in series rather than appearing randomly.
Compare the Sharpe ratio only with Sharpe ratios calculated on the same basis. Values from other software often use different return intervals and annualization, so the numbers are not interchangeable.
Drawdown fields
MT5 reports drawdown three ways, each for balance and for equity:
- Absolute: how far the account fell below the initial deposit.
- Maximal: the largest fall in deposit currency from a local high to the next low, with the percentage in brackets.
- Relative: the largest fall in percentage terms from a local high to the next low, with the money amount in brackets.
Maximal and relative drawdown can come from different episodes. Suppose a $10,000 test rises to $11,000, falls to $9,900, later climbs to $20,000 and falls to $18,500:
| Field | Value | Episode |
|---|---|---|
| Balance Drawdown Absolute | $100 | Lowest balance below the $10,000 deposit |
| Balance Drawdown Maximal | $1,500 (7.50%) | $20,000 to $18,500 |
| Balance Drawdown Relative | 10.00% ($1,100) | $11,000 to $9,900 |
The largest money loss came late, when the account was bigger. The largest percentage loss came early. For risk planning, relative drawdown is usually the more useful number, because it scales with the account.
Equity drawdown includes floating losses on open positions; balance drawdown only changes when trades close. A grid or averaging EA can show a modest balance drawdown and a far larger equity drawdown, because it carries losing positions until they recover. If the two differ sharply, trust the equity figure. The maximum drawdown guide explains why the recovery math gets harsh as drawdowns deepen.
Trade counts and streaks
- Total Trades: deals that closed a position and fixed a profit or loss.
- Total Deals: every deal, including entries. A simple in-and-out strategy shows roughly twice as many deals as trades.
- Short Trades (won %) and Long Trades (won %): counts and win rates by direction. A large imbalance can reveal a strategy that only worked because the test period trended one way.
- Profit Trades and Loss Trades (% of total): the overall win and loss counts.
- Largest and Average profit and loss trade.
- Maximum consecutive wins ($) and Maximum consecutive losses ($): the longest streak by count, with its money total.
- Maximal consecutive profit (count) and Maximal consecutive loss (count): the most expensive streak by money, with its count.
The last two pairs look alike but answer different questions. The longest losing streak is not necessarily the most costly one. Use the longest streak to test your drawdown limits and the costliest one to test your risk per trade.
Common misreadings
- Reading results before settings. A test on generated ticks with no delay is a best case for short-term systems.
- Taking net profit at face value. Check lot sizing and compounding settings first.
- Ignoring the trade count. Fewer than a few hundred trades leaves wide error bands on every ratio.
- Using balance drawdown. Equity drawdown is closer to what you would have lived through.
- Treating an optimized result as a test. The best pass from an optimization is in-sample by definition. Its out-of-sample record is the evidence, as discussed in the guide to backtest overfitting.
To analyze the trades outside MT5, save the tester's trade table as CSV and load it into the trade log analyzer, which recalculates profit factor, win rate and drawdown and adds sample-size and cost-sensitivity checks. A good backtest is a reason to keep testing, not evidence that a strategy will make money live.
Key points
- Check the modeling mode, History Quality and delay setting before reading any result.
- Expected payoff is average money per trade; compare it with the amount risked, not with other EAs.
- MT5 reports absolute, maximal and relative drawdown for balance and equity; relative equity drawdown is usually the most useful.
- Longest losing streak and costliest losing streak are separate fields with separate uses.
- Treat optimized results as in-sample and look for out-of-sample evidence.