Same EA, same dates, two results: 6 Strategy Tester settings that decide your backtest

18 November 2026, 10:30
Fernando Henrique Funicelli
0
10

Run an EA twice in the Strategy Tester, on the same dates, changing only the tick generation mode. The result changes, sometimes a lot.

The EA is the same. The prices it was fed are not. Inside a 1-minute bar, gold can touch both the stop and the target, and only the real tick knows which came first.

Six choices on the Settings tab decide whether your test measures the EA or flatters it. The rule behind all six: a backtest is only as real as the prices it uses.

The six settings, side by side

SettingThe choice that flattersThe honest choice
Tick generation mode"1 minute OHLC" or "Open prices only" on an EA that trades in minutes"Every tick based on real ticks" for the test that decides
DatesThe good year, without the hard oneThe recent period to decide, the full history to find the largest drop
Initial deposit and leverageA deposit or leverage you don't have: the test's lot is not your lotThe ones you will really use
Execution"No Delay", an execution no live account hasA delay, such as "Random Delay"
SymbolAnother broker's data, an account without your commissionYour broker and your account type; a demo account there works
Optimization and ForwardTuning until the curve is perfectFew parameters, Forward on, suspicion of the perfect curve

And one thing to read, not to set: equity, which also counts what is still open. Equity sinking below the balance is a losing position being held, so judge the test by the largest drop on the equity line.

Tick generation mode: what each mode feeds the EA

It decides which prices the tester uses to simulate each trade. On an EA that trades fast, no other field moves the result as much.

ModeWhat the tester usesWhen it fits
Every tick based on real ticksEvery price the broker recordedThe test that decides, above all for fast EAs
Every tickTicks generated from the 1-minute barsWhen real ticks are missing
1 minute OHLCOnly the open, high, low and close of each minuteA quick screen of an idea
Open prices onlyOnly the opening price of each barEAs that decide at the bar's open
Math calculationsNo pricesCalculations, not trading

With example prices: a 1-minute bar opens at 4,000.00, makes a high of 4,003.00 and a low of 3,997.50, and closes at 3,998.80, below the open. You bought at 4,000.00, with the stop at 3,998.00 and the target at 4,002.50, both inside the bar. "1 minute OHLC" knows four points and has to guess their order. "Every tick" builds a path, and on a bearish bar like this one it goes to the high first: the test books the target and the profit. Only the real tick shows that, here, the stop came first.

On an EA that trades in minutes, that doubt repeats over hundreds of trades. That is how a fast test invents profit that never existed.

Dates: the same year can give two drops

2025 and 2026 decide: they are the market the EA meets tomorrow. The older years show the worst case. One example of ours, in Pitbull Gold's test with real ticks from January 2021 to October 2026: the largest continuous drop was 49%, in 2024. Test 2022 on its own and it reaches 59%. Two ways of testing, two drops; size the account by the larger one.

Forward: the slice the optimization never saw

Optimizing runs the EA with hundreds of parameter combinations and keeps the best. Before you know the rule works, that only teaches the EA to memorize the past. The first test runs with optimization off and Forward on "No".

When you do optimize, Forward holds back the end of the period (1/2, 1/3, 1/4, or "Custom" from a date) and shows that slice in the Forward tab. The optimization never sees it. If the result collapses there, the optimization memorized the past.

BacktestForward, in the testerDemo account, live
Where it runsThe tester, on historyThe tester, on an end slice the optimization never sawYour broker, in real time
What it provesThe rule repeats in the pastThe rule was not tuned only to the pastReal execution: spread, slippage, hours, a terminal that stays on
When to use itAlways, firstWhen there was optimizationBefore real money

The full step-by-step guide is on our site, with the numbered tester screen, how to read the result, figures and FAQ: how to backtest on MT5 with real ticks.

Rerun our three tests with these settings

Each of our three EAs has a free demo for the Strategy Tester and its test sheet on the page: real ticks, real spread, real commission.

EAPeriodDepositBroker and symbolLeverageDefault profile
Collie GoldJanuary 2021 to October 22, 20261,000 USDVantage, XAUUSD+1:1000Moderate
Pitbull GoldJanuary 2021 to October 31, 202610,000 USDVantage, XAUUSD+1:500Balanced
Beagle GoldJanuary 2023 to October 202610,000 USDVantage, XAUUSD+1:100Moderate

Each page's year-by-year table restarts every year with the same deposit, so rerun one year at a time. One number from each, in the test with real ticks:

  • Collie Gold on the Market: +586% in 2025 on the Moderate profile. Largest continuous drop: 37%.
  • Pitbull Gold on the Market: +884% in 2025 on the Balanced profile, and 21 of 22 months in profit in 2025 and 2026. Largest continuous drop: 49%, in 2024.
  • Beagle Gold on the Market: +184% in 2025 on the Moderate profile, and 21 of 22 months in profit in 2025 and 2026. Largest continuous drop: 27%.

Then switch to your own conditions: your deposit, your broker, 2025 and 2026. All three recognize gold under any symbol name and detect the broker's time zone on their own, so choosing the profile is the whole setup.

Lucky Dogs Trading Systems