Gold Breakout Regime
- Experts
- Versione: 1.0
- Attivazioni: 10
Gold Breakout Regime - a gold breakout EA fitted to the 2023-2026 gold uptrend,
which says so in its first line and switches itself off when that uptrend ends.
No grid, no martingale, no averaging down, a hard stop on every trade. XAUUSD,
any timeframe (the breakout is on M15), MetaTrader 5.
You decide one number: how far down you are willing to let the account go. Lot
size, entry, stop, target and when to stop trading - the EA works those out.
== WHAT IT DOES ==
It watches the high of the last 96 M15 bars - 24 hours - and when a bar's high and
close are both above it, buys at the open of the next bar. Stop 15 USD below
entry, or the recent swing low if closer; target 1.5 times the stop. One trade at
a time, no trailing stop and no time exit, so a trade can run over a week (longest
11.8 days), weekends included. XAUUSD and M15 are recommendations, not
restrictions.
== YOU CHOOSE THE DRAWDOWN ==
You set Max drawdown budget; the default is 30%. The EA divides it by the worst
drawdown the strategy recorded in the fit window, in R-multiples (19.79), giving
your risk per trade: 1.51592% of equity at 30%. The calculation is printed at
startup and on the panel.
budget risk/trade arithmetic min fit-window DD fit-window CAGR
20% 1.01061% 1,484.25 USD 20.37% 19.06%
30% 1.51592% 989.50 USD 29.47% 29.39%
40% 2.00000% 750.00 USD 37.60% 38.60%
And it is not decoration: at 30% the EA stops opening trades once equity is 30%
below its peak, which over the whole record happened twice, in 2013 and 2018.
But the budget caps one episode, not your lifetime: the EA waits 30 days,
re-anchors the peak to current equity and starts again, so the account steps down
more than once - over the whole record a 30% budget realised 55.83%, a 20% budget
39.59%. The recovery resume fired zero times in twenty years. A 40% budget is
clamped to the 2.0% ceiling in code, and the guard watches account equity, so
other EAs on it move it.
== THE FIT WINDOW: WHAT WE OPTIMISED, AND WHAT IT SCORED ==
The lookback, stop and target were chosen by measuring 72 configurations on
January 2023 - August 2026, so the figures below were measured on the data that
selected them: an in-sample result. XAUUSD alone, 30% budget, compounded, the
defaults as shipped, Model 1 (ticks from M1 bars).
668 trades over 3.63 years, 184 a year. CAGR 29.4%, 2.55x
Profit factor 1.187 in R-multiples, 1.130 in account currency, 1.149 ex-swap
Maximum drawdown 29.5% (the relative drawdown, the larger of the two MT5
definitions). Longest flat period 334 days
Win rate 44.6%, average win / average loss 1.40, longest losing streak 9
The hard stop never fired inside that window (29.47% against a 30.00% budget), so
none of it is what the safety net produced. Where the money came from: 2023 gave
+8.0% of the fit-window profit, 2024 +14.0%, 2025 +85.6%, 2026 to August -7.6%.
Taking 2025 out drops the profit factor to 1.0858; removing the ten best trades
leaves 1.147 - one year of gold behaving as fitted, not a few lucky trades. Nor
does it end on a high: equity peaked 3 February 2026 and the last day is 24.1%
below that, so you would start inside a drawdown.
== THE WHOLE RECORD, APRIL 2006 - AUGUST 2026 ==
1,712 trades, 84 a year. Profit factor 1.108, CAGR 9.35%, 6.18x
Maximum drawdown 55.83%, longest losing streak 18
4,639 days - 12.70 years - without a new equity high
13 of 21 years positive. Worst year 2018 -21.54%, best 2025 +98.99%
By regime: 2006-2012 rising PF 1.2468; 2013-2015 falling PF 0.6104, -55,639 USD;
2016-2022 sideways PF 1.0629; 2023-2026 fit window PF 1.1909. Five of the 17
rolling five-year windows are below 1.05, four below 1.00, worst 0.7786. And "184
a year" is a fit-window figure: the whole record is 84, 2013-2015 was 19. Every
year and every window is in the risk disclosure.
== THE REGIME FILTER, AND THE ARITHMETIC WE ARE NOT HIDING ==
The EA measures gold's own 12-month change, 252 daily bars. Below -10% it stops
opening new positions, above 0% it starts again; open positions are never
touched. On by default, replaying the whole history at startup.
filter ON (default) filter OFF
Net profit +513,413 USD +1,003,094 USD
Profit factor 1.108 1.124
Maximum drawdown 55.83% 54.92%
In 2013-2015, when gold fell, it cut the loss by 59%: -38.93R without it, -15.92R
with it. Over the whole record it costs nearly half the profit and does not
reduce the drawdown - it saved 23.01R in that fall and gave back 64.92R in the
three rising markets, a net -41.91R over 20.39 years. So we will not tell you the
filter makes this EA profitable. It does not. It is insurance, and an input you
can switch off - with it off nothing stands between this strategy and the next
multi-year decline. We ship it on. It reads XAUUSD; if yours is XAUUSD.m, point
Regime symbol at it.
== REAL TICKS, SWAP, AND THE ACCOUNT YOU NEED ==
Real ticks exist only for August 2024 - August 2026, about two years, and 51 whole
days are missing inside it: 90.35% tick coverage. "100% real ticks" is not
something we can write. That window on Model 4 gives 482 trades,
1.166 and a 23.4% drawdown; on Model 1, 510 trades, 1.118 and 29.5%. Those numbers
are better and we are not selling them to you: 482 trades in the strongest part of
the fitted trend is not evidence.
A long gold position pays swap every night and this EA is long only: swap is
-11.8% of net profit in the fit window and -24.9% over the whole record. Strip it
out and the fit-window profit factor rises to 1.149, so this strategy is not
flattered by carry - but the cost is not comfortable, and MT5 applies today's rate
to all of it.
Recommended 1,500 USD, minimum 1,000 USD; below that this EA does not trade, and
the drop is a cliff, not a slope. Of 668 fit-window trades, 500 USD takes 1
(0.1%), 800 USD takes 15 (2.2%), 1,000 USD takes 642 (96.1%), 1,500 USD all 668.
The arithmetic minimum is 989.50 USD and we will not print it without this table:
97.9% of the stops are exactly 15 USD, so nearly every signal
needs the same account. At a 20% budget it rises to 1,484.25 USD. The 15 USD stop
is a gold number: on XAGUSD the same code would need 49,475 USD and take nothing.
== WHAT IS BUILT IN, AND BEFORE YOU BUY ==
Built in: a hard stop on every position, sent with the order and verified after;
the 1.5R target; sizing capped at 2.0% per trade in code; the drawdown hard stop
and its cool-off; the regime filter; spread and volume-limit protection; daily
loss and profit limits that watch this EA's own trades on its own symbol, not
your account as a whole; a News filter, off by default (the tester cannot reach
the MT5 calendar, and every figure here used no news filter); a 15-line panel; and
31 inputs in 7 groups.
Live, the EA resolves time through GMT; in the tester it falls back to a
daylight-saving model (winter GMT+2, US rules). Both are printed at startup and on
the panel - check them against your broker's clock. If you reproduce our backtest,
avoid Wednesday: MT5 applies the swap rate live when you press Start to all twenty
years, and Wednesday's rollover is triple; ours were measured on Friday 4
September 2026. Every number here is backtest data: there is no live track record,
no MQL5 signal and no reviews. We run this EA on our own live account
(since 2026-09-04).
== TESTING, OUR OTHER GOLD EA, AND PRICE ==
Behind this one EA: three self-checks on the research code, 12 grid points, and 72
fit-window configurations, of which 71 were discarded. The EA was then matched
against the research code trade by trade - 781 entries, every stop distance
identical, all 20 regime switch dates matching to the day - then 22 confirmation
runs over 21,868 trades and the 32-case validation simulation, all 32 trading.
Across this catalogue we have thrown away roughly 78,000 configurations; four
became products, and we publish the rejections.
Bullion Trend Follower is our defensive gold EA: H4 trends on gold and silver with
an ATR trailing exit, validated across the whole 20.4-year record rather than a
fitted window. That one was tested everywhere; this one was fitted.
Leveraged trading carries a substantial risk of loss; past results do not indicate
future results. The user guide, risk disclosure and three .set files are in the
Comments tab, downloadable by anyone. Updates are free.
The launch price is the lowest this EA will ever sell for: it rises one step
per ten copies sold and stops permanently at 199 USD - a ceiling, not
a target. A stated limit is checkable.
