EA strategy check: backtest red flag checker

What this check does, and what it cannot do

A strategy check reads the statistics an Expert Advisor already published or that your own MetaTrader 5 Strategy Tester produced, and asks the questions an experienced buyer asks before paying: is the sample big enough, is the profit worth the drawdown, does the test include costs, does a stop protect every trade, and is there any live proof. Each answer is a pass, a note, a warning or a red flag, and the score is simply 100 minus a penalty for each. It counts warning signs. It does not forecast profit, and a high score is not a promise. The same questions are behind our scored EA verdicts, where a human reads the sources behind the numbers.

Where to find each number in MT5

Run the test in the Strategy Tester, open the Results tab and read the table. Profit factor, Total trades, Profit trades (% of total), Average profit trade, Average loss trade and Equity drawdown maximal are all printed there. Net profit as a percentage is the Total net profit divided by your starting deposit. Use the equity drawdown, not the balance drawdown, because equity includes the losses a basket is carrying while it waits. For the modelling, check the Tester settings: "Every tick based on real ticks" is the only mode that uses your broker's actual tick history. Our guide to reading an MT5 backtest report walks through every field with a worked example: how to read an MT5 backtest report.

Why these thresholds

Trades and years. Under about 100 trades, one lucky week can account for most of the profit, so a profit factor of 2 tells you almost nothing. A strategy also needs to meet more than one kind of market. Three years that include a trend, a range and a shock is a very different test from one calm year.

Profit factor. Below about 1.3 the edge is thinner than the spread and slippage you will pay live. Above about 3.5 it is more often a sign of curve-fitting, a hidden grid that has not yet met its bad run, or a test that left out costs than of a genuinely better strategy.

Drawdown and recovery factor. The recovery factor is net profit divided by the worst drawdown. Below 1 the strategy earned less than it risked. Drawdown also compounds against you: a 25% fall needs a 33% gain to recover, and a 50% fall needs 100%. The drawdown in a backtest is a floor, not a ceiling; live drawdowns are usually deeper than tested ones.

Win rate against payoff. Win rate alone says nothing. A strategy that wins 90% of trades and loses ten times as much on the other 10% has zero expectancy before costs. The check multiplies the two to get expectancy per trade, and warns when a very high hit rate sits on a tiny payoff, which is the usual shape of an EA that looks perfect until one large loss arrives.

Grid and martingale. These designs win often and lose rarely but heavily. With no stop the backtest has simply not met the move that ends the account. Our deeper reads: martingale strategy risks and grid trading explained.

Four ways a good-looking backtest lies

1. Costs left out. Spread, commission and slippage cost the same on every trade, so they hurt the strategies that trade most. A scalper that makes 0.8 pips a trade before costs may make nothing after them.

2. Open-price modelling. This mode only looks at the first price of each bar. It cannot see whether the stop or the target was reached first inside the bar, so a short-hold strategy gets flattering, unreliable results.

3. Optimising on everything. If the settings were tuned across the whole test, the result shows how well the settings fit the past. Only data the optimiser never saw can say anything about the future. See how to optimise an EA without overfitting and analysing a backtest with QuantAnalyzer.

4. One test, no live record. A backtest is a claim. A verified live record on one account, with deposits and withdrawals visible, is evidence. Our track record checklist shows how to tell the difference.

After the check

A low score means the statistics you have are not enough to trust, not that the EA is a scam. Ask the vendor for the missing pieces, rerun the test on real ticks with costs, or open a demo account first. To see how independently scored EAs compare, use the EA screener. If you want a tested starting point, browse the Expert Advisors in our store; each listing shows what ships with it, so you can run this same check on the vendor's numbers.