Cross Pair Forge for cTrader: three markets with a prop-firm guard

What it does

Cross Pair Forge trades three markets from one chart - AUDJPY, NZDCAD and gold (XAUUSD) - with a prop-firm guard built in. You choose your firm's programme, the phase and the account size; the guard loads that programme's limits, the strategy sizes every trade from them, and the robot stops before a limit is reached. Each market has one tactic. On gold, an Asian-range breakout: the high and low of the 5-minute bars from 00:00 to 06:00 UTC mark the range, a 5-minute close beyond it between 06:00 and 11:00 UTC opens the trade, the stop sits at the middle of the range and the target at 1.5 times the risk, and a trade still open at 20:00 UTC is closed. On AUDJPY, a four-hour compression breakout: a four-hour bar narrower than each of the six before it arms the tactic, and a 15-minute close beyond that bar opens a trade only in the direction of the four-hour 50 EMA, with the target at two times the risk and a trailing stop from one times the risk. On NZDCAD, a momentum shift: a 5-minute RSI extreme (above 81 or below 19) that turns back, taken only while the one-hour ADX is at least 33, and never a buy inside a supply zone or a sell inside a demand zone; half of the position closes at one times its risk, the stop then moves to break-even and a trailing stop follows the rest. At most one position is open per symbol, and the risk of all open trades together is capped at three times the risk per trade. Each trade gets its stop loss and take profit with the broker right after the fill, and every trade is closed on Friday at 20:00 UTC. There is no grid, no martingale and no averaging down.

With the Strategy preset on Auto (the shipped default and the safe challenge mode), the risk per trade comes from the room your rules leave: the maximum loss, less when it trails your equity, divided by ten and rounded down to a fixed step between 0.25 and 3 %. A funded account trades 0.5 % at most, with no increase after profits. The guard checks the whole account on every price change of any of the three symbols: it measures the daily loss the way the firm does, on the firm's own clock, follows static and trailing maximum-loss floors, counts the target only after the minimum trading days, applies consistency rules, profit and open-loss caps, news and weekend rules where the firm has them, and refuses an entry whose full stop would no longer fit inside the room that is left. A panel on the chart shows the firm, the phase, the firm's terms, the risk, the three symbols and the open risk.

The firm's rules and terms

The Rules list holds 39 programme profiles of the 9 firms whose published terms allow a purchased robot on cTrader, phase by phase, and a Custom rules group for any other firm. Where a firm's terms require it, the robot never holds a buy and a sell together, holds one position at a time, keeps each trade open for the firm's minimum time before closing it itself, caps the risk per trade or per day, keeps the risk constant, or counts a trading day only when a trade opens and closes on it; the panel shows which of these apply. The 24 firms whose terms do not allow this robot are in the list too: choosing one stops it at start with the firm's own words. A robot you buy takes the same trades for every buyer, and several firms flag identical trades across accounts as copy or group trading. Cross Pair Forge does not disguise its trades; read your firm's rules on this before you buy a challenge.

The published backtest

The figures on this page are cTrader's own backtest report of the shipped defaults: a two-step programme's funded phase (5 % daily loss, 10 % static maximum loss), Strategy preset Auto at 0.50 % risk per trade. The run used cTrader's backtester on IC Markets real tick data, on an AUDJPY chart trading AUDJPY, NZDCAD and XAUUSD, 100,000 USD, leverage 1:100, commission 27.5 USD per million USD of volume, from 2 January 2026 to 5 October 2026. Every number is read from the run's own report file: net profit 39,635.86 USD, a return of 39.64 %, profit factor 1.64, 253 deals of which 147 won (a partial close counts as a deal of its own), a maximum balance drawdown of 3.77 % and a maximum equity drawdown of 4.02 %, with nine of ten months closing in profit. The same settings on a 10,000 USD account gave +34.67 % with a profit factor of 1.63.

What that backtest cannot promise. Auto was chosen on 2026 prices, so this shows how it traded the year it was chosen on; it is not a forecast. In quieter markets it trades less and earns less: on the deals closed from January to May the profit factor was 2.16, on those closed from June to October 1.25, and the first days of October closed at a loss. The robot never rounds a lot up: below the broker's 0.01-lot minimum an entry is skipped - on 10,000 USD the same settings opened 225 of the 232 entries - which is why the recommended balance is 10,000 USD. The result of any run depends on its period, broker data, balance and leverage; run your own backtest on your own broker's data before trading.

What the challenge simulation shows

Auto and Fast were simulated on every firm challenge phase they trade, on real 2026 ticks of the three symbols, started from a fresh account on every weekday and judged by the firm's exact limits; each figure is the average over the phases. Auto lost no account in any group of starts. Of its January and February starts, 20 % passed within 14 days, 70 % within 30 days and 99 % within 120 days; of the March to May starts, 68 % within 30 days and 93 % within 120. Of the 1 to 5 June starts, made after the configuration was chosen, 7 % passed within 30 days and 84 % within 120. Fast passed 89 to 90 % within 30 days on the January to May starts, but it lost 1 % of the January and February starts and 21 % of the June starts. On funded accounts, where both modes trade 0.5 % at most with no increase after profits, none of 4,107 starts on 37 funded programmes was lost within 120 days. A pass within 30 days is therefore not reliable across market regimes, the configuration was chosen after the January to May months had been seen, and an account can be lost. On the ten challenge phases where Fast lost an account in its test, Fast refuses to trade and asks for Auto.

How it was verified

The engine is platform-free C#: it references no cTrader type, so the code that ships can be compiled outside the platform and replayed on the same ticks cTrader's backtester uses, the three symbols in cTrader's own order. Replayed that way at release, nine cases - the published settings among them, and Auto and Fast chosen through the Strategy preset on challenge phases - matched cTrader's own reports deal by deal, 1,000 deals, to the cent, with zero mismatches, and the simulator behind the challenge figures matched the same reports just as exactly. The release build, its debug-log twin and the development build produced identical reports of the published settings. The published figures are cTrader's own, not the replay's.

Presets

Everything is inside the cBot and chosen with the Strategy preset parameter. Auto, the shipped default, is the safe challenge mode described above. Fast is the opt-in mode for a quicker challenge: it divides the room by six instead of ten, adds 0.25 % for every 2 % of closed profit, up to 1 % more, and refuses the ten challenge phases where it lost an account in its 2026 test. A funded account trades 0.5 % at most in both modes, with no increase after profits. Custom engine opens every setting of all 14 tactics: the three above and eleven more (band reversion, an RSI pullback, an efficiency-ratio regime switch, currency strength, liquidity sweeps, short-term extremes, and five models with a confluence score: trend pullback, Bollinger-Keltner squeeze, z-score reversion, SuperTrend and Ichimoku), each assignable to any of up to four symbols; its defaults are the tested three-market portfolio at 0.5 % risk. There is no file to download, find or import after the purchase.

Requirements

cTrader Desktop 5.x on Windows or Mac, or cTrader Web, on a hedging or a netting account - it holds at most one position per symbol. One instance on any chart trades the three symbols, found by your broker's spelling of AUDJPY, NZDCAD and XAUUSD; a symbol the account does not offer is skipped. It builds its own 5-minute to daily bars on any chart timeframe. The recommended balance is 10,000 USD. It declares no access rights at all; the news calendar is read through cTrader's own networking service. 363 parameters, every one documented, with a prop-firm guide and the firm rules phase by phase. Keep the platform running, use a VPS, or run it in cTrader's cloud; on another computer the guard rebuilds its progress from the account's own trade history.

Support

Through the Discussion tab of the Store listing, so the conversation stays next to the product. Please include your broker, cTrader version, the firm and phase you chose, your parameter values and the instance log.