How Much Do Backtests Differ by Broker? One EA, 4 Brokers, 10 Years
Contents
Bottom line: net profit differs several-fold, but most of the gap is compounding
The same EA, with the same settings over the same period, produced 10-year compounding backtests whose net profit differed several-fold, and in one case more than tenfold, between brokers. Measured as "profit ÷ lots" per closed trade, however, GOLD VIPER stays within 0.97–1.21x on all four brokers. Most of the gap is a small early difference amplified by compounding.
The several-fold gap is the result for the two gold EAs (GOLD VIPER and GOLD NEURON). MEGAMAX DONCHIAN · OVERDRIVE on USDJPY differed by only about 1.3 times.
Two other causes mattered. The first is server time: on a GMT+0 broker the EA trades at different moments. The second is price data: even the same trade fills at slightly different prices on each broker. Maximum drawdown changed much less between brokers than net profit did.
🎁 Download Free EA
Recommended Broker Comparison →📧 Free 8-email course + subscriber-only pack of 13 MT5 indicators
Right after you sign up, you get a download link for a subscriber-only pack of our 13 MT5 indicators: compiled files (.ex5), source code and a setup guide. Then 8 emails on algo-trading essentials, reading backtests correctly and choosing a broker (one a day for the first 5 days, then 7, 14 and 30 days after signing up). You also get notified before each price rise of the versions sold on the MQL5 Market (launch pricing that steps up as copies sell). Our EAs are free on this site with a live account at a partner broker.
* Privacy strictly protected. You can unsubscribe at any time.
Why we tested this
Most buyers backtest on MetaTrader 5's default server (MetaQuotes-Demo) and then trade on their own broker's account. Brokers differ in price data, spreads, server time and swaps, so the same EA produces different results. We measured how large that difference is with three of our EAs.
Test conditions
| Item | Setting |
|---|---|
| Period | 2016.10.01 – 2026.10.01 (10 years) |
| Model | Every tick (real ticks where available) |
| Deposit / leverage | $10,000 / 1:500 |
| EA | Identical ex5 file and identical inputs (standard settings) on every broker |
| Money management | Compounding (lot size grows with balance) |
| Broker | Server | Symbols | Notes |
|---|---|---|---|
| MetaQuotes | MetaQuotes-Demo | XAUUSD / USDJPY | MT5 default. Server time GMT+2/+3 |
| Exness | Exness (Standard account) | XAUUSDm / USDJPYm | Server time GMT+0 |
| XM | XMTrading-MT5 3 | GOLD / USDJPY | Real gold ticks start 2025.05.08 |
| HFM | HFMarketsGlobal-Demo4 | XAUUSD / USDJPY | XAUUSD: Price history starts in 2017 |
"History quality" is the share of real ticks the tester could use. The lower it is, the more ticks the tester generated. All figures were measured as of 2026-10-31.
Results
GOLD VIPER v2.41 (XAUUSD)
| Broker | Net profit | PF | Trades | Win rate | Max DD (largest peak-to-trough, equity) | Max relative DD (equity) | Swap | Quality |
|---|---|---|---|---|---|---|---|---|
| MetaQuotes | +$1,638,130 | 1.49 | 12,418 | 56.7% | 29.5% | 32.7% | −$34,632 | 77% |
| Exness | +$4,433,046 | 1.62 | 13,149 | 56.6% | 24.4% | 30.7% | −$236,200 | 7% |
| XM | +$347,903 | 1.41 | 12,233 | 55.6% | 23.4% | 29.7% | −$48,125 | 13% |
| HFM | +$798,433 | 1.53 | 10,321 | 57.3% | 26.4% | 32.4% | −$76,672 | 2% |
By year (return on start-of-year balance / max balance drawdown within the year)
| Year | MetaQuotes | Exness | XM | HFM |
|---|---|---|---|---|
| 2016 (from Oct) | +63.6% / 14.1% | +38.8% / 6.8% | +49.8% / 12.3% | no history |
| 2017 | +8.3% / 19.9% | +48.1% / 15.6% | −6.7% / 17.3% | 0.0% / 7.0% |
| 2018 | +38.0% / 24.6% | +70.5% / 24.4% | +22.8% / 19.7% | +25.2% / 19.9% |
| 2019 | +44.4% / 14.7% | +103% / 10.6% | +60.0% / 12.3% | +53.5% / 14.3% |
| 2020 | +132% / 20.3% | +134% / 10.1% | +101% / 15.6% | +238% / 13.7% |
| 2021 | +89.2% / 13.7% | +104% / 15.9% | +48.7% / 12.3% | +72.7% / 16.6% |
| 2022 | +21.5% / 17.9% | +22.6% / 15.2% | +4.4% / 18.9% | +14.7% / 17.5% |
| 2023 | +112% / 28.5% | +134% / 19.5% | +36.5% / 25.7% | +53.6% / 27.6% |
| 2024 | +110% / 9.5% | +114% / 11.0% | +103% / 11.0% | +108% / 13.8% |
| 2025 | +68.9% / 10.9% | +56.9% / 14.8% | +23.3% / 11.2% | +28.9% / 13.9% |
| 2026 (to Sep) | +16.2% / 22.4% | +35.7% / 19.9% | +22.3% / 18.6% | +52.4% / 20.4% |
GOLD NEURON v3.19 (XAUUSD)
| Broker | Net profit | PF | Trades | Win rate | Max DD (largest peak-to-trough, equity) | Max relative DD (equity) | Swap | Quality |
|---|---|---|---|---|---|---|---|---|
| MetaQuotes | +$20,822 | 1.30 | 1,157 | 54.3% | 14.7% | 23.8% | −$1,144 | 77% |
| Exness | +$97,607 | 1.61 | 1,181 | 56.8% | 15.9% | 24.9% | −$7,835 | 7% |
| XM | +$12,457 | 1.22 | 1,180 | 53.6% | 30.8% | 30.8% | −$3,725 | 13% |
| HFM | +$16,412 | 1.26 | 1,019 | 54.3% | 30.4% | 30.4% | −$4,013 | 2% |
By year (return / max balance drawdown within the year)
| Year | MetaQuotes | Exness | XM | HFM |
|---|---|---|---|---|
| 2016 (from Oct) | +14.5% / 6.7% | no trades | +15.7% / 6.9% | no history |
| 2017 | −17.9% / 22.0% | +5.4% / 10.2% | −20.0% / 20.2% | +2.3% / 0.0% |
| 2018 | +7.3% / 16.9% | +8.1% / 22.8% | −0.8% / 18.2% | +1.7% / 17.5% |
| 2019 | +8.6% / 11.1% | +24.9% / 13.5% | +1.5% / 12.0% | +2.2% / 11.8% |
| 2020 | +30.3% / 5.9% | +60.5% / 6.7% | +18.9% / 6.5% | +49.6% / 4.2% |
| 2021 | +23.8% / 9.6% | +38.4% / 9.1% | +18.9% / 8.4% | +20.6% / 11.0% |
| 2022 | −0.9% / 12.3% | +10.5% / 8.5% | +2.5% / 8.9% | −3.9% / 12.4% |
| 2023 | +10.4% / 17.4% | +22.5% / 12.1% | +1.7% / 19.1% | −9.3% / 25.2% |
| 2024 | +33.9% / 7.0% | +74.2% / 8.4% | +45.0% / 8.8% | +48.8% / 7.0% |
| 2025 | +21.9% / 3.9% | +42.6% / 4.3% | +12.8% / 4.5% | +5.1% / 7.5% |
| 2026 (to Sep) | −2.3% / 11.4% | +1.2% / 11.9% | −0.1% / 11.2% | +1.0% / 13.5% |
MEGAMAX DONCHIAN · OVERDRIVE (USDJPY)
OVERDRIVE USDJPY was merged into MEGAMAX DONCHIAN in October 2026. The test used the former OVERDRIVE USDJPY ex5, which places the same trades as MEGAMAX DONCHIAN set to SL_ATR 2.5, Trail_ATR 4.0 and RiskPercent 3.0 (checked on MetaQuotes). A 3% risk is an aggressive setting.
| Broker | Net profit | PF | Trades | Win rate | Max DD (largest peak-to-trough, equity) | Max relative DD (equity) | Swap | Quality |
|---|---|---|---|---|---|---|---|---|
| MetaQuotes | +$16,372 | 1.28 | 722 | 41.3% | 14.6% | 19.5% | −$53 | 100% |
| Exness | +$13,044 | 1.25 | 691 | 39.9% | 20.4% | 20.4% | $0 | 7% |
| XM | +$14,672 | 1.26 | 719 | 41.0% | 14.0% | 20.2% | +$1,061 | n/a |
| HFM | +$13,976 | 1.25 | 718 | 40.5% | 19.4% | 19.6% | $0 | 4% |
By year (return / max balance drawdown within the year)
| Year | MetaQuotes | Exness | XM | HFM |
|---|---|---|---|---|
| 2016 (from Oct) | +10.4% / 3.7% | −0.9% / 0.9% | +10.9% / 3.6% | +10.7% / 3.7% |
| 2017 | −3.2% / 8.2% | −3.8% / 9.9% | −3.0% / 8.7% | −4.4% / 9.0% |
| 2018 | −0.9% / 7.0% | −2.2% / 7.1% | −1.0% / 6.7% | −1.9% / 8.0% |
| 2019 | −0.1% / 8.8% | +0.8% / 10.2% | +1.5% / 8.7% | +1.5% / 8.3% |
| 2020 | −6.6% / 16.1% | −4.8% / 16.1% | −7.1% / 16.5% | −6.4% / 15.8% |
| 2021 | +41.5% / 3.0% | +46.8% / 3.0% | +37.9% / 3.8% | +37.5% / 3.3% |
| 2022 | +40.9% / 7.2% | +41.5% / 6.2% | +41.9% / 7.5% | +48.7% / 6.6% |
| 2023 | +8.6% / 5.5% | +7.2% / 6.1% | +8.4% / 6.3% | +8.1% / 6.2% |
| 2024 | +14.0% / 7.4% | +16.8% / 7.6% | +7.4% / 9.9% | +10.2% / 9.2% |
| 2025 | +19.2% / 4.6% | +17.4% / 6.7% | +18.0% / 7.1% | +17.0% / 6.7% |
| 2026 (to Sep) | −9.3% / 13.1% | −15.6% / 19.0% | −8.6% / 12.4% | −14.5% / 18.0% |
Where the differences come from
1. Compounding turns small differences into large ones
All three EAs size positions from the balance, so a small early difference snowballs over 10 years. To remove the compounding effect we summed "profit ÷ lots" over every closed trade (profit per lot).
| EA | MetaQuotes | Exness | XM | HFM |
|---|---|---|---|---|
| GOLD VIPER | 2,210,321 | 2,680,469 (1.21x) | 2,146,162 (0.97x) | 2,366,143 (1.10x vs MetaQuotes from 2017) |
| GOLD NEURON | 107,977 | 209,137 (1.94x) | 91,764 (0.85x) | 101,196 (0.98x vs MetaQuotes from 2017) |
| MEGAMAX DONCHIAN · OVERDRIVE | 24,934 | 17,219 (0.86x vs MetaQuotes from 2017-03) | 22,943 (0.92x) | 22,522 (0.90x) |
HFM's price history only starts in 2017, so it is compared with MetaQuotes from 2017 onward. GOLD VIPER's net profit differs about 13x between the highest (Exness) and the lowest (XM), yet per lot all four brokers are within 0.97–1.21x.
2. Server time (GMT offset) changes when the EA trades
MetaQuotes, XM and HFM all run on GMT+2/+3. Exness runs on GMT+0, so daily candles and time-of-day filters shift by 2–3 hours and the EA trades at different moments. Of GOLD NEURON's 108 entries in 2025, 87 on XM and 86 on HFM were at the same time as on MetaQuotes. On Exness only 67 matched even after shifting by 2–3 hours, so about 40% were different trades. That is why GOLD NEURON's profit per lot is roughly double on Exness alone.
3. Price data (history quality and spread)
Even for the same trade at the same time, the fill price differs between brokers. On XM, GOLD NEURON earned 15% less per lot than on MetaQuotes with almost the same trades; that gap is the accumulated difference in fill prices from spread and generated ticks. Exness, XM and HFM keep only a short history of real gold ticks (quality 2–13%), so most of the 10 years is filled with ticks generated by the tester.
On USDJPY, MEGAMAX DONCHIAN · OVERDRIVE moved far less. From March 2017 on, of MetaQuotes' 695 entries, 650 on XM, 645 on HFM and 639 on Exness (after shifting by 2 to 3 hours) matched within 30 minutes. Four data notes: Exness USDJPYm placed only two very small trades before March 2017, both at exactly 00:00, which suggests thin price history, so its per-lot figure is compared from March 2017. XM's report shows history quality as n/a. HFM has USDJPY history from 2016, so it is compared over the full 10 years. Swap was zero on Exness and HFM.
Checking on your own broker
- Backtest on your own broker's server as well. MetaQuotes-Demo results are a reference point. On a broker whose server time is not GMT+2/+3, even the trade timing changes.
- Look at history quality. Periods with low quality are results on generated ticks. Re-checking only the period with real ticks makes the broker difference clearer.
- Compare per lot and drawdown, not net profit. Compounded net profit swings widely with early differences. Size your capital assuming a maximum relative drawdown of around 30%.
- Our broker comparison is on the broker page.
Summary
- With the same EA and settings, a 10-year compounding backtest can differ several-fold, even more than tenfold, in net profit depending on the broker
- Most of that gap is compounding: per lot, GOLD VIPER stays within 0.97–1.21x and GOLD NEURON within 0.85–1.94x
- On a GMT+0 broker (Exness) the EA trades at different times altogether
- Maximum relative drawdown changed far less than net profit: 30–33% for GOLD VIPER and 24–31% for GOLD NEURON
- On USDJPY the gap was small: MEGAMAX DONCHIAN · OVERDRIVE made +$13,044 to +$16,372, about 1.3 times, with maximum relative drawdown of 19.5–20.4%
- See GOLD VIPER, GOLD NEURON and MEGAMAX DONCHIAN for each EA, and the measured ranking to compare all our EAs
- For the general causes see why the same EA performs differently across brokers, and for the time shift see server time zone and session filter drift
Backtest results are based on past data and do not guarantee future profits.
Related
2026-10-27
Demo vs Live Spread Gap: Estimating Backtest Shrinkage
2026-10-25
What Equity Curve R² Tells You: Doubting Too-Smooth EAs
2026-10-24
Always Check the Worst Month in Your Backtest
2026-10-22
Same Trade Count, Different Reliability: Independence
📧 Free 8-email course + subscriber-only pack of 13 MT5 indicators
Right after you sign up, you get a download link for a subscriber-only pack of our 13 MT5 indicators: compiled files (.ex5), source code and a setup guide. Then 8 emails on algo-trading essentials, reading backtests correctly and choosing a broker (one a day for the first 5 days, then 7, 14 and 30 days after signing up). You also get notified before each price rise of the versions sold on the MQL5 Market (launch pricing that steps up as copies sell). Our EAs are free on this site with a live account at a partner broker.
* Privacy strictly protected. You can unsubscribe at any time.
Have a question about EAs or FX?
You can post it on the Nattzy trading forum (free sign-up). Questions beyond this article are welcome too — EA settings, choosing a broker and more.
Ask on Nattzy ↗