AI Algorithmic Trading Solution

FxMath Elite Pro — GBPUSD H1 Optimised Preset: +12.31%, 90.2% Win Rate, Out-of-Sample Confirmed

FxMath Financial Solution

FxMath Elite Pro — GBPUSD H1 Optimised Preset: +12.31%, 90.2% Win Rate, Out-of-Sample Confirmed

The optimiser evaluated 2,904 configurations across 7,913 GBPUSD H1 bars and confirmed the winner on an untouched out-of-sample slice. The result: +12.31% net return, a 90.2% win rate, a 4.89 profit factor — and a maximum drawdown of just 2.51%.

FxMath Elite Pro GBPUSD H1 optimised preset — 4.89 profit factor, 90.2% win rate, +12.31% return, 2.51% max drawdown
FxMath Elite Pro — GBPUSD H1 optimised preset at a glance.
Executive Summary

377 trades · 90.2% win rate (340W / 37L) · +$1,231.20 (+12.31%) · Profit Factor 4.89 · Max Drawdown 2.51% · all 16 months positive · out-of-sample +4.17%

Headline Performance

+$1,231.20
Net Profit · +12.31%
90.2%
Win Rate · 340W / 37L
4.89
Profit Factor
2.51%
Max Drawdown
377
Trades
$513.92
Minimum Account
1.80
Sharpe · Sortino 1.68
2
Worst Losing Streak

Equity Curve

From a $10,000 starting deposit the account climbed steadily to $11,231.20. Buying and holding GBPUSD over the same window returned just +0.78% — the strategy does not depend on a sustained trend.

GBPUSD H1 optimised preset equity curve
Figure 1: Cumulative equity — $10,000 deposit to $11,231.20, versus buy & hold (+0.78%).

Walk-Forward Validation

In-sample metrics are an upper bound by construction. The honest number is the untouched out-of-sample slice — and it stayed profitable.

Metric In-sample Out-of-sample Note
Net profit $807.55 $416.50
Return % 8.08% 4.17% OOS must stay positive
Max drawdown % 2.51% 2.34%
Profit factor 4.89 4.90
Trades 249 124 OOS 2026-02-27 to 2026-08-28
Win rate 90.4% 90.3%
Sharpe 1.78 1.99
Exposure 100% 100%

Monthly Consistency

Every one of the sixteen months was positive — best month +$135.10 (July 2025), weakest +$26.80 (May 2025).

GBPUSD H1 monthly P&L
Figure 2: Monthly P&L in USD — all 16 months positive.
Month P&L Month P&L
2025-05 +26.80 2026-01 +27.80
2025-06 +55.40 2026-02 +134.50
2025-07 +135.10 2026-03 +109.20
2025-08 +53.60 2026-04 +54.70
2025-09 +108.70 2026-05 +81.40
2025-10 +82.40 2026-06 +54.70
2025-11 +81.50 2026-07 +108.30
2025-12 +81.50 2026-08 +35.60

Same Engine — Real Tuning

Out of the box the EA made $2.39 on 2 trades. With the optimised GBPUSD H1 preset it made $1,231.20 on 377 trades. The engine was always capable; the inputs were not calibrated for this symbol.

EA defaults versus optimised preset
Figure 3: Defaults versus the optimised preset — net profit, profit factor, trades and win rate.
Metric EA Defaults Optimised
Net profit $2.39 $1,231.20
Return 0.02% 12.31%
Max drawdown 0.02% 2.51%
Profit factor 1.04 4.89
Trades 2 377
Win rate 50.0% 90.2%

Shipped Settings — GBPUSD_H1.set

  • Grid: no stop loss (0) · take profit 280 pips · grid distance 112 pips · max 8 levels · fixed 0.01 lots
  • Labels: 30 pip stop / 75 pip target
  • Forest: 20 trees · max depth 8 · min samples 10 · class weight 2 · trained on 2,000 bars, retrained every 48 bars
  • Regime: adaptation ON · trend filter OFF
  • Protection: enabled · 30% max drawdown · stop-on-protect · trend lot multiplier 0.2

Minimum Capital — $513.92

The binding constraint is the drawdown budget, not margin. One 0.01-lot position costs about $2.68 of margin; the peak stack of 0.24 lots needed $64.27. But margin is not released while a position is underwater, and the worst floating loss was $265.55 — so the true floor is $329.82. The $513.92 figure is 2.0 × the $256.96 worst backtest drawdown; that headroom is what makes the strategy survivable rather than merely funded.

How the Engine Works

  • 4 Random Forests (80 trees) trained on 120 features directly inside MetaTrader 4 and 5
  • Online retraining every N bars — the model is never stale
  • Regime-aware adaptive grid — range, weak trend and strong trend modes
  • 4-model consensus — Buy, Sell, Trend and Hedge must agree
  • Bank-grade protection — drawdown halt, strong-trend auto-exit, stop-on-protect
  • Zero dependencies — no DLLs, no Python, no external data; pure MQL

Pricing & Payment

Licenses are currently 50% off. Every plan runs on up to 3 accounts and ships both the compiled .ex5 (MT5) and .ex4 (MT4) files.

1 Month
$69
full access · 3 accounts

Buy Now

6 Months
$199
save $215

Buy Now

12 Months
$299
save $529 · most popular

Buy Now

24 Months
$399
save $1,257

Buy Now

Best Value

Lifetime
$799
one-time · all future updates

Buy Now

Payment: cryptocurrency only — BTC · ETH · USDT · BNB · SOL · LTC · DOGE and more. After payment you receive the compiled .ex5 and .ex4 files instantly.

Get the optimised preset and the full report at elite.fxmath.com/reports/gbpusd-h1. Watch live results on the dashboard and get free realtime signals in the Telegram channel t.me/FxMathAI. Support: @FxMath.

Risk Warning: This is a grid strategy with no stop loss at 100% exposure, so the downside is unbounded in principle. A sustained trend can exceed the drawdown budget sized on this 19-month sample. Trading forex, metals, indices and cryptocurrencies involves substantial risk of loss. The figures above are backtest and out-of-sample results, not verified live results, and are not indicative of future performance. Only risk capital you can afford to lose. Nothing here is trading advice.

Tags

FxMathElite ProGBPUSDForexOptimised PresetMetaTraderMT5MT4Machine LearningRandom ForestGrid TradingAlgorithmic Trading


Discover more from FxMath Financial Solution

Subscribe to get the latest posts sent to your email.

Follow me on Twitter

About FxMath

FxMath Financial Solution is a financial software team developing end-to-end algo trading systems for quantitative hedge funds and institutional trading groups. Our system based on profitable mathematical models with highest possible profit and lowest drawn down.

Subscription

Subscribe to Trade Signals via Email

Enter your email address to subscribe to our free trading signals by email

Discover more from FxMath Financial Solution

Subscribe now to keep reading and get access to the full archive.

Continue reading