A gold grid without a stop loss, 2008 to 2026.
Grid EAs win almost every trade and their charts look beautiful. We ran a typical gold grid through 18 years of real prices to see what the pretty chart hides.
The short answer
A typical gold grid without a stop loss made about 5% a year over 18 years, less than simply buying gold and holding it. On the way it lost 69% of the account in 2013. At twice the trade size, the account was wiped out.
$1,000 in each, 2008 to 2026
Account value with open trades counted at their lowest point each hour. Trading costs and the overnight fee (swap) included.
Grid EA: +160%. Gold held: +399%. Grid at twice the size: account wiped out on 26 June 2013.
The idea
A grid EA buys gold, buys more every time the price falls a bit further, and closes all trades at a small profit when the price comes back up. It wins almost every time, so the account curve looks beautiful.
The danger: there is no stop loss, no automatic exit when things go wrong. If gold falls far and stays down, the loss keeps growing.
The rules we tested
- Buy one unit. Every time gold falls another 0.65%, buy one more.
- The units slowly get bigger: 1, 1, 1, 2, 2, 2, 3, 3, 3 and so on.
- When all open trades together are in a small profit, close them all and start again.
- No stop loss. Size: 0.01 lot for every 500 of money in the account.
Tested on gold prices for every hour from January 2008 to September 2026, with the spread and the swap included.
What happened
- Total profit 2008 to 2026: +160%, about 5.2% a year.
- Worst fall: −69% of the account, in June 2013, when gold kept falling for months.
- At the worst moment 55 units were open at once, and one group of trades stayed open for 245 days.
- Simply buying gold and holding it made +399% in the same time, with a worst fall of −45%.
Bigger size means wiped out
- Twice the size: the account was wiped out on 26 June 2013.
- Three times the size: wiped out on 9 September 2008.
- Five times the size: wiped out on 12 August 2008.
Limiting the number of trades does not fix it. With at most 8 units the grid made +13% in 18 years, still fell 34%, and one group of trades stayed stuck for more than seven years.
The start date changes everything
Started in any year from 2008 to 2013, the grid went through the same −68% to −69% fall in 2013. Started in 2014 or later, the worst fall was "only" −29% to −32%. A backtest that starts in 2020 has never seen a year like 2013.
So before you run any grid, ask for a test that includes 2008 and 2013, at the size you will actually trade.
What it means for you
- A grid without a stop loss does not avoid losses. It hides them in open trades until one long fall takes everything at once.
- The trade size decides whether you survive. The same rules at twice the size ended with an empty account.
- A high win rate says nothing about risk. Look at the deepest fall, and at how long trades stay open.
Where Giza comes in
Giza Gold EA is built the other way round: no grid, no martingale, a stop loss on every trade, and every trade sized from a fixed % of the account, so a losing trade costs what you planned. Every trade it closes is shown on our live results page, losses included.
Every trade, in public.
Watch Giza Gold EA trade gold and stock indices, losses included, updated every 5 minutes.