The short answer
We simulated gold grid systems and a zone-recovery hedge on real hourly COMEX gold prices from 30 October 2024 to 30 October 2026, each with a $10,000 account, 1 ounce (0.01 lot) per grid level, 1:100 leverage and a stop-out at a 50% margin level.
- Every grid was stopped out over the two years: buy, sell and two-way, at $10 and $25 spacing.
- The $10 buy grid looked superb for 15 months: 3,825 winning trades and $37,103 banked. Then gold fell from its January 2026 peak and the account was stopped out on 2 February 2026 with 95 positions open.
- The $10 sell grid lasted 148 days against gold’s 2025 rally; it was stopped out in every one of 36 six-month windows.
- A zone-recovery hedge ($10 zone, $20 target) completed 41 profitable cycles and was stopped out after 33 days, with 645 ounces open in a single cycle.
These systems win almost every trade until the market trends. Gold trends a lot.
How a gold grid works
| Grid type | Opens | Closes | Stop loss |
|---|---|---|---|
| Buy grid | A buy each time price falls through a level (every $10, say) | Each buy at one level higher | None |
| Sell grid | A sell each time price rises through a level | Each sell one level lower | None |
| Two-way (“hedged”) grid | Both | Both | None |
In a sideways market, price oscillates through the levels and every oscillation closes a small winner. The trap is what happens in a trend: each level passed adds another losing position, and the losses add up like a staircase. A $500 move against a $10 grid leaves 50 open positions with an average loss of about $250 each, roughly $12,500 of floating loss on a $10,000 account.
Simulation set-up
Prices. COMEX gold futures (GC=F) 1-hour bars, 30 October 2024 to 30 October 2026; gold started the window at $2,794. Within each bar we assume the usual path (up bar: open, low, high, close; down bar: open, high, low, close) and process every grid level crossed along it. Account. $10,000, 1 ounce per position, $0.30 per ounce cost on each closed trade, margin at 1:100 (price / 100 per ounce), stop-out when equity falls to 50% of used margin. For two-way grids, margin is charged on the larger side only, as many brokers do for hedged positions, which is generous. Not included: swaps (overnight financing), which would make every result worse because these systems hold dozens of positions for weeks.
Grid results: full two-year run from 30 October 2024
| Grid | Stopped out on | Days survived | Price at stop-out | Winning trades closed before | Profit banked before | Open positions at stop-out |
|---|---|---|---|---|---|---|
| Buy, $10 | 2 Feb 2026 | 460 | $4,677 | 3,825 | $37,103 | 95 |
| Sell, $10 | 28 Mar 2025 | 148 | $3,122 | 664 | $6,441 | 58 |
| Two-way, $10 | 16 Apr 2025 | 168 | $3,344 | 2,378 | $23,067 | 80 |
| Buy, $25 | 23 Mar 2026 | 508 | $4,322 | 1,031 | $25,466 | 53 |
| Sell, $25 | 16 Apr 2025 | 168 | $3,357 | 158 | $3,903 | 33 |
| Two-way, $25 | 22 Apr 2025 | 173 | $3,486 | 349 | $8,620 | 38 |
Read the “profit banked” column carefully. It is real, closed profit, and it is exactly what a grid robot’s trade history shows. The floating loss on the open positions was larger, and at the stop-out the broker closed everything. The $10 buy grid’s equity at stop-out was about $2,205, after it had banked $37,103.
Grid results: rolling six-month windows
To avoid depending on one start date, we started each grid every 10 days and ran it for six months (36 windows).
| Grid | Windows stopped out | Median days to stop-out | Survivors’ median ending equity |
|---|---|---|---|
| Buy, $10 | 18 of 36 (50%) | 58 | $21,719 |
| Sell, $10 | 36 of 36 (100%) | 56 | — |
| Two-way, $10 | 34 of 36 (94%) | 64 | $6,439 |
| Buy, $25 | 15 of 36 (42%) | 63 | $15,117 |
| Sell, $25 | 26 of 36 (72%) | 97 | $11,899 |
| Two-way, $25 | 27 of 36 (75%) | 98 | $12,884 |
The buy grids did best simply because gold rose for most of the period; their survivors made a lot. That is not an edge, it is a leveraged bet on direction without a stop. In a falling year, the table flips.
How zone-recovery “hedging” works
Zone recovery is sold as a way to never close at a loss. The cycle:
- Buy 1 ounce at price P. Take profit at P + $20.
- If price falls to P − $10 (the zone), sell a larger position, sized so that reaching P − $30 closes the whole group at a profit.
- If price returns to P, buy a still larger position, sized so that reaching P + $20 closes everything at a profit.
- Repeat until one take-profit is hit.
The sizes this requires, with a $10 zone and $20 target, starting at 1 ounce:
| Flip | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 |
|---|---|---|---|---|---|---|---|---|---|
| New position (oz) | 1 | 2.5 | 3.75 | 5.62 | 8.44 | 12.66 | 18.98 | 28.48 | 42.71 |
Each new position is about 1.5 times the last, but the total open grows much faster because every earlier position stays open. After 12 flips the group holds hundreds of ounces.
Zone-recovery results (same account)
| Set-up | Outcome on the full run | Cycles completed | Largest cycle | Six-month windows stopped out |
|---|---|---|---|---|
| $10 zone, $20 target | Stopped out 2 Dec 2024, day 33 | 41 | 13 positions, 645 oz open | 36 of 36 |
| $20 zone, $40 target | Stopped out 2 Dec 2024, day 33 | 11 | 13 positions, 645 oz open | 33 of 36 |
| $5 zone, $10 target | Survived to 30 Oct 2026, equity $51,680 | 5,227 | 14 positions, 969 oz open | 12 of 36 |
The $5 version deserves an honest note: on the full run it survived and multiplied the account five times. But at its worst it held 969 ounces at once on a $10,000 starting balance, and a third of the six-month windows failed. It survived because of where it happened to start, not because the method is safe.
Hedging a single losing trade
Outside of robots, many traders “hedge” a losing gold trade by opening an equal position in the opposite direction. Consider a 1-lot (100 oz) long at $4,050 that has fallen to $4,020:
| Choice | Position | Locked loss | Extra costs | What you still have to decide |
|---|---|---|---|---|
| Close the long | Flat | $3,000 | None | Nothing |
| Hedge with a 1-lot short | Long + short | $3,000 (fixed while both are open) | A second spread, swaps on two positions | When to lift each leg, which is two decisions instead of one |
The hedge does not reduce the loss; it freezes it and adds cost. Removing one leg later is simply a new directional trade. Our page on margin calls and stop-outs explains how hedged positions still consume margin at many brokers.
What the simulations say, in plain English
- High win rates are a design feature, not evidence of skill. The $10 buy grid closed 3,825 winners before a single realised loss, and then lost the account.
- Grids are directional bets in disguise. A buy grid is long gold with no stop; a sell grid is short gold with no stop. The direction of the market decides survival.
- Two-way grids are not neutral in a trend. One side banks small wins while the other accumulates large losers.
- Recovery hedges are martingale with extra steps. Position size grows geometrically, exactly as in our martingale simulation.
- Our results are flattering. No swaps, no widened spreads at stop-out, generous margin on hedges.
Red flags in grid and hedge robots
- No stop loss in the strategy description, or “the system manages losses”.
- Many positions open at once in the trade history, at evenly spaced prices.
- Equity much lower than balance for weeks at a time.
- A backtest that starts conveniently at the bottom of a range.
The EA checklist lists the questions to ask any seller.
What to do instead
Every trade should have a stop and a size calculated from it. Check how much margin a position really uses with the gold margin calculator, size it with the position size calculator, and keep total open risk within a fixed limit. Our risk management lesson sets out the rules.
Check it yourself
If you run or are considering a grid robot, export its trade history and plot equity, not balance, over time. Then ask: how far would gold have to move against the grid to hit stop-out? Over the two years we tested, gold rose about $3,085 from its low ($2,541.50 on 14 November 2024) to its high ($5,626.80 on 28 January 2026), then fell about $1,671 to $3,955.40 by 30 June 2026.
Learn it with other traders
Grid and recovery systems are hard to judge from their marketing. In the XAUUSD Global community, members dissect robot results and share honest equity curves. The first month is free, then $19.99 a month.
Educational content only, not financial advice. Trading gold with leverage carries a high risk of loss.
Frequently asked questions
What is a grid trading strategy on gold?
A grid places trades at fixed price intervals, for example every $10. A buy grid buys each time price falls through a level and takes profit one level higher; a sell grid does the opposite; a two-way grid does both. There is usually no stop loss, so losing positions stay open until price comes back.
Why do grid strategies blow up gold accounts?
Because gold trends. In a long move against the grid, losing positions pile up at every level and the floating loss grows roughly with the square of the distance travelled. In our simulation, a $10 buy grid started at $2,794 banked $37,103 over 15 months, then was stopped out on 2 February 2026 with 95 positions open after gold fell from its peak.
Is a two-way (hedged) grid safer?
Not in our test. A two-way $10 grid was stopped out after 168 days and failed in 34 of 36 six-month windows. When gold trends, the side against the trend accumulates losers faster than the other side banks small wins.
What is zone recovery hedging?
A system that opens an opposite, larger position when a trade moves a set distance against it, sized so that reaching either take-profit closes the whole group at a small profit. Sizes grow with each flip: with a $10 zone and $20 target, about 1, 2.5, 3.75, 5.6, 8.4, 12.7 ounces and so on.
Does hedging a losing gold trade reduce the loss?
No. Holding a long and an equal short locks the loss at its current size, adds a second spread and two sets of swaps, and leaves you needing to decide when to remove each leg. It is economically the same as closing the trade, but with extra cost.
Did any grid or hedge survive the simulation?
A zone-recovery set-up with a tight $5 zone survived the full two-year run, but its largest cycle needed 969 ounces open at once, and a third of six-month windows ended in a stop-out. Every fixed grid we ran was eventually stopped out.
Sources and further reading
- CME Group — Gold futures overview — Contract specifications and trading hours for COMEX gold
- CFTC — Commitments of Traders — Weekly futures positioning, including gold
- Investor.gov — Stop order — How stop orders work, and their limits in fast markets
External sites are independent of XAUUSD Global. Check dates and data yourself.