XAU Squeeze Sniper 【BASE】｜Gold-specialized "Explosion Sniper" Breakout EA

Development History

“Recently gold prices have been so volatile that normal EAs can’t withstand the drawdown…”

“But if you use a stability-focused EA, you’ll miss big trends…”

Do you have such worries?

With inflation, geopolitical risks, and the surge of algorithmic trading, modern markets are rapidly becoming highly volatile.

Old EAs designed for the calm markets of the past (10 years ago) cannot survive the coming rough waters.

January 30, 2026. Gold moved 768 dollars in one day.

“Spread widened,” “stop-loss didn’t trigger in time,” “stopped out at dawn”—these voices flooded social media.

In 2024, gold rose by +26% for the year. Average daily range was about $33.
In 2025 +65%. Average $60 per day. Still, people said it was getting rough.

And in 2026.
January +13.33%. February +10.58%. March -16.56%.
Average daily range is $170—the fivefold of 2024.

In this market, what would happen if you ran averaging down or martingale?
Just imagining makes your spine go cold.

Right now, best-selling products at GogoJungle are

EA that claim “AI-equipped” and “win rate above ●●%” dominate the top in the 30k–50k yen range.
Sales themselves aren’t bad. If there’s demand, they sell—this is market principle.

However, reading those reviews reveals another view.

“Forward tests win cleanly, but entries don’t occur on my real account”
“I don’t understand why AI entered here. If I lose several in a row, I can’t decide whether to keep running.”
“Using averaging down expands unrealized losses and I can’t sleep at night. But it’s stable, so I can’t let go.”

What users really want is not flashy numbers but the security of being able to verify why it wins or loses themselves.
And as many major EAs remain MT4-only, users want MT5 support.

So we developed

【XAU Squeeze Sniper】.

This system specializes in pinpointing the “explosive trends” unique to gold (XAUUSD) with Squeeze (squeeze break) & Trend logic*.*FULL version only

Additionally, with an original “Hibernate Filter (Regime Detection)” derived from ten years of backtest analysis, it achieves overwhelming defense: aim for explosive gains in favorable markets, and physically stop the system in losing markets.

XAU Squeeze Sniper is a “defense-first” designed, explosive-snipe.

Maximum drawdown 6.08%. After 10 years 3 months and 1,572 ticks of full tick backtests, this is the result.
No averaging down. No martingale. Maximum position 1.
Not “win big”but“don’t exit.”

Why this design—behind the verification data

This EA was not “picked by intuition.”

First, compare both “contrarian (mean-reversion)” and “trend-following” with data.
Backtest results for contrarian inside the band: best PF 0.89.
Squeeze breakout (trend-following): PF 1.26–1.46.
The numbers were clear. The conclusion to stop contrarian wasn’t intuition, but actual results from running the system.

All backtests were conducted on real broker data. Not MetaQuotes demo (ideal spread), but history in actual market environments. A filter that automatically skips entries when spreads exceed 80 points prevents wasted trades during widening spread periods.

And,

Comprehensive search across 200+ AI parameter combinations.
With Bollinger Bands period × deviation × squeeze-detection window × ATR stop loss multiplier × risk-reward × various filters—the total combinations exceed 200 when run through automated scripts. Even if fast tests show good numbers, only those that survive full tick-by-tick (history quality 98%) precise testing are used. It wasn’t about picking good numbers by guess; it was about testing all and selecting the best.

Three core themes

① Trend-following logic that discarded contrarian—data-backed choice

This EA does not perform contrarian trading at all.
The decision to not do it wasn’t based on feel. After implementing and testing the contrarian module, data confirmed that it underperforms trend-following in the mid-to-long term for gold.

This EA is a trend-following strategy that detects breakouts from Bollinger Band squeezes (constriction) and rides the initial movement of the trend.
Specific entry, exit, and filter thresholds are disclosed in detail in the logic manual for purchasers. You can verify why entries were made or passed on with your own eyes.

Stop loss:

• Dynamically set via ATR(○○*) × SL multiplier* disclosed in manual
• SL width automatically adjusts with market volatility, reducing unnatural stop losses in rough markets like 2026

Take profit:

• Close at a fixed value by applying risk-reward ratio to SL distance (RR=○○*)* disclosed in manual
• When a trend forms, profits are allowed to run with a “big win, small loss” mechanism

② Five-layer defense—“Hibernate filter” and four safety devices

The biggest enemy of breakout strategies is the back-and-forth in low-volatility range markets.
This EA constantly monitors the higher-timeframe ADX. A period of low volatility is deemed a “hibernate period,” and entries are physically blocked.

• Hibernate filter (Winter filter): Do not enter when daily ADX is below 25 in low-volatility markets. The most crucial safety valve to prevent false breakouts in range markets. This single feature quietly carried us through the low-volatility hell from 2016–2021
• Time-of-day filter: automatically avoids times with statistically many false signals
• Day-of-week filter: stop entries on Thursdays (to avoid post-midweek reversal patterns)
• Weekend entry stop: automatically stop new entries after Friday 8 PM
• Daily guardian: if daily loss reaches 5% of free margin, suspend all trades for that day. The EA automatically decides the exit, preventing emotional mistakes

“A defensive EA” may look dull, but surviving traders are almost always dull.

③ AI parameter optimization: selecting from 200+ combinations

The default values of this EA are not decided by gut feeling.

BB period (20–35) × BB deviation (1.8–2.5) × squeeze-detection window (5–20) yields 64 BB grid combinations scanned quickly.
Then adding ATR stop loss multiplier × risk-reward ratio, Trend module parameter sweeps, and filter cost-benefit analyses pushes total tested combinations beyond 200.

Final verification on all ticks (history quality 98%) shows a PF deviation from 1M OHLC of ≤ 0.01. Even if fast tests look good, only those surviving precise tests are used.

Set on XAUUSD M15 chart, this selected configuration moves as-is.

Backtest results (all ticks / history quality 98%)

Long-term 10 years 3 months OOS test (robustness across unknown periods including 2016–2021 low-volatility):

Note: 2016–2021 gold markets were range-bound with low volatility. The Hibernate Filter reduced wasted entries during this period, so explosive returns like in recent years were not produced. Maintaining a PF ≥ 1.0 through 10-year OOS testing suggests genuine edge, not curve-fitting.

Recent 4 years 3 months In-Sample (optimization period):

Note: Backtests were conducted in ECN-real environments. Spreads are applied to execution prices, and trading costs ($0.72 per trade) are added separately. After deducting all real costs, the 10-year PF remains 1.09. In the last two years (2025–2026), gold moved $60–$170 per day, a market fit for this EA to maximize its potential.All real costs including spreads and fees are accounted for, and PF remains 1.09 over 10 years. The last two years (2025–2026) have seen explosive gold movement averaging $60–$170 per day, a market stage where this EA can fully demonstrate its power.

Numbers show Gold’s “abnormality”

2024 daily was $33. 2026 daily is $170. It’s a different kind of gold movement.
In this environment, breakout strategies show their true value. At the same time,overly forgiving EAs get eliminated in an instant.

Q&A

Q. Do you use averaging down or martingale?
A. Not at all. We exclude logic that increases a losing position.

Q. Will it operate with brokers that have wide spreads?
A. If spreads exceed 80 points, entries are automatically skipped (InpMaxSpreadPoints setting). All backtests are in ECN real environments, reflecting real costs including spread and commissions. Broker choice directly affects PF. Choosing ECN/STP accounts with lower costs is the simplest way to raise PF long-term.

Q. Will it run with any broker?
A. It will run with any broker that offers MetaTrader 5.

Q. What margin do you recommend?
A. Start from $1,000 (about 160,000 yen) with 0.01 lot. The BASE version has a low max drawdown of 6.08%, so you can start with small amounts safely.

Q. Should I change the parameters?
A. Generally, use default values. The defaults were chosen after scanning over 200 combinations, balancing PF, DD, and trade count most stably. However, for those who want customization, there is a paid option (Parameter Adjustment Guide) explaining three profiles—defensive, balanced, and aggressive—based on real backtest data.

Who this EA fits

• Have experienced losing funds with averaging-down EAs
• Want to prioritize survival over win rate
• Want to exploit gold’s explosiveness but don’t have time for discretion
• Dislike black-box EA internals—want to understand why entries were made
• Looking for a MT5-compatible gold EA

Who this EA does not fit

• Hope to enter every day (this EA follows a “wait” logic)
• Choose by win rate
• Believe backtest guarantees future results

For those who want even more aggressiveness —XAU Squeeze Sniper FULL

BASE version is a defensive model with 6.08% drawdown.
Meanwhile,XAU Squeeze Sniper FULLunlocks a pyramiding feature to add positions when a trend is present.

In the same 10-year backtest, net profit rose from $1,826 to $6,797—about 3.7x (not including fees).
Risk increases, but you can choose between offense and defense yourself.

First, confirm the EA’s robustness with the BASE version.
If satisfied, move to the next stage—