Same EA. Three brokers.
We pointed the same blade we used on 800+ market EAs at our own lineup. All 24 EAs, run on XMTrading, HFM and Exness real ticks — identical window, identical settings, shipping defaults. Does the same EA behave the same when the broker changes? Here is every number, including the ones we would rather not show.
Why
Why run the same EA on different brokers
Even a real-tick test proves little if it comes from a single broker — the result may simply fit that one feed. Spreads, fill behaviour, server timezone, symbol naming and contract specs all change from one broker to the next. An EA that draws a beautiful curve at one firm can drain the account at another. So we run the same binary with the same settings at several brokers at once and compare.
Test conditions
- Scope
- All 24 EAs in the published lineup
- Brokers
- XMTrading / HFMarkets / Exness demo accounts
- Ticks
- Model 4 real ticks requested; actual coverage XM 100% / HFM 54–56% (see note)
- Window
- 2026.01.02 – 2026.07.31 (identical across brokers)
- Capital
- $10,000 / leverage 1:500
- Settings
- Shipping defaults, untouched
Result
21 of 24 kept the same sign across brokers
The table compares HFM against XM. 21 EAs kept the same sign of P&L at both brokers, 2 flipped, and exactly one broke at a single broker. Equity-drawdown differences stay within a few points for almost every symbol — the behaviour of the EA itself reproduces across brokers.
| EA | HFM P&L | XM P&L | HFM equity DD | XM equity DD | Trades (HFM/XM) |
|---|---|---|---|---|---|
| GOLD NEURON | +2,781,698 | +1,696,133 | 16.38% | 21.14% | 447 / 444 |
| BITCOIN METEOR | +1,971 | +1,588 | 36.21% | 36.67% | 10 / 10 |
| BITCOIN NOVA | +1,874 | +1,476 | 30.43% | 30.60% | 10 / 10 |
| TRINITY TREND | +886 | +331 | 14.73% | 18.35% | 38 / 38 |
| NASDAQ DIP BUYER | +331 | +746 | 2.98% | 6.23% | 2 / 9 |
| MONTHLY BURST | +492 | +675 | 9.17% | 9.40% | 55 / 53 |
| BITCOIN GLACIER | +380 | +306 | 4.39% | 4.62% | 3 / 3 |
| BITCOIN COMET | +233 | +264 | 12.27% | 12.27% | 12 / 12 |
| SOLANA COMET | +22 | +190 | 6.77% | 6.43% | 11 / 11 |
| BITCOIN NANO | +157 | +111 | 2.72% | 2.88% | 3 / 3 |
| GOLD BLITZ | +74 | +75 | 3.81% | 3.79% | 5 / 5 |
| ETHEREUM TREND | +24 | +4 | 10.48% | 10.97% | 8 / 8 |
| AUREUS GOLDsign flipped | +75 | -233 | 20.51% | 20.69% | 18 / 19 |
| GOLD PIVOTsign flipped | -553 | +300 | 10.07% | 5.86% | 24 / 24 |
| MEGAMAX USDJPY | -199 | -328 | 11.56% | 12.51% | 56 / 60 |
| GOLD VIPER | -289 | -764 | 23.93% | 24.54% | 18 / 19 |
| BITCOIN COIL | -386 | -401 | 7.38% | 7.43% | 3 / 3 |
| ATLAS PORTFOLIO | -459 | -1,004 | 26.75% | 28.26% | 154 / 173 |
| MEGAMAX GBPJPY | -797 | -974 | 22.37% | 23.77% | 119 / 121 |
| MEGAMAX EURJPY | -807 | -969 | 22.42% | 23.61% | 119 / 121 |
| ETHEREUM NOVA | -1,068 | -1,372 | 27.70% | 29.04% | 7 / 7 |
| OVERDRIVE USDJPY | -1,541 | -1,931 | 31.63% | 33.88% | 56 / 57 |
| DONCHIAN ENGINE | -199 | -328 | 11.56% | 12.51% | 56 / 60 |
| GOLD KING v6defect found | +6,270 | -9,772 | 54.72% | 98.61% | 345 / 451 |
Note: DONCHIAN ENGINE (the free engine) runs the same core on the same symbol as MEGAMAX USDJPY, so the identical figures are expected, not a typo.
Please do not read this table as a performance record
This is seven months, default settings, demo accounts — not a profitability claim. Several EAs traded fewer than ten times, so their P&L carries no statistical weight. The only thing to read here is whether the same EA behaved the same way across firms. Each EA's actual multi-year real-tick record lives on its own page.
What broke
One EA broke — at one broker only
GOLD KING v6 returned +6,270 at HFM but -9,772 at XM: 97.7% of the account gone, equity drawdown 98.61%. Same binary, same settings. The cause turned out to be a defect in the EA. It measured the offset between server time and GMT only once at startup, so after a daylight-saving switch in March or November it kept running one hour out. For a strategy that enters on session times, a one-hour shift destroys the strategy itself.
We changed detection to re-measure every week from the Monday opening gap, and re-ran the same window on the shipping binary (v6.21).
| Before fix | After fix (shipping v6.21) | |
|---|---|---|
| Net profit | -9,772 USD (-97.7%) | +7,760 USD (+77.6%) |
| Profit factor | 0.48 | 3.11 |
| Max equity drawdown | 98.61% | 64.18% |
| Trades | 451 | 390 |
Testing at a single broker would never have surfaced this. The pre-fix binary was profitable at both HFM and Exness. This is exactly what running across several brokers buys you.
Full disclosure
What we will say plainly
The two brokers differ in real-tick coverage
Asking MT5 for Model 4 (real ticks) only fills the periods the broker actually distributes; MT5 synthesises the rest from M1 bars. Measured coverage here was 100% at XM but 54–56% at HFM — so roughly half of the HFM column rests on generated ticks. The XM-versus-HFM gap therefore reflects tick quality as well as broker behaviour. That is the same objection we raise against market EAs, so the same note belongs on our own table. Every MT5 report states its own "History Quality: N% real ticks" — always read that line before you buy anything.
Even after the fix, equity DD is a high 64%
GOLD KING v6 shows only 4.16% balance drawdown, while equity drawdown — which counts open, unrealised losses — reaches 64%. That gap is typical of grid-structured EAs, and it is why you should be suspicious whenever a seller shows you balance drawdown alone. It is also why we always publish the equity figure.
Two EAs flipped sign
AUREUS GOLD and GOLD PIVOT changed the sign of their P&L between HFM and XM. Both swings sit within ±5% of starting capital, which over seven months and roughly twenty trades we read as noise. We publish them anyway, as concrete evidence that broker choice moves the result.
Many EAs are negative in this window
Cut out the specific seven months of January–July 2026 and nearly half of the 24 sit slightly negative. Most are low-frequency trend followers that simply did not get their setup and traded a handful of times. That a short window cannot rank EAs is precisely the argument we use against market EAs — so we apply the same ruler to ourselves.
About the data
Every figure here is extracted mechanically from the actual reports emitted by the MT5 Strategy Tester. The reports are kept in our internal archive and are cross-checked against the live figures whenever a published number changes. We will share individual reports on request.
The inconvenient numbers go out first
We wrote that we real-tick tested 800+ market EAs and that most were phantoms. That criticism only stands if we hold ourselves to the same ruler. A defect in our flagship, the fix, the re-verification, and the weakness that still remains — all of it goes here.
Disclaimer: the figures on this page describe past results over a fixed period and do not promise future performance. Automated trading carries the risk of losing your capital. All values were measured between 2 January and 31 July 2026 with $10,000 of starting capital, 1:500 leverage and shipping default settings.