Does Martingale Work in Forex?

The question, and what a fair answer looks like

"Does martingale work in forex?" has two honest answers, and they do not conflict. The ladder does what it is built to do: most cycles end with a small win at an early step. It also has a cost that arrives rarely and is large. Whether that adds up to "working" depends on how large and how rare, and that cannot be answered without the numbers for a specific ladder on a specific pair. So this page does not give a verdict. It shows the arithmetic with every cost counted and lets you see where it breaks.

For the bot form and the idea of a step cap, see martingale bot. To try your own numbers, use the martingale calculator. For a related method that sizes recovery trades from the zone rather than by doubling, see zone recovery.

What forex adds: spread, swap and margin

• Spread. The difference between the buy and the sell price. It is paid when a trade opens, so every step starts slightly behind. A stop gets hit a little sooner than the chart suggests, and a target is reached a little later. See spread.
• Swap. A charge or credit for holding a position past the daily rollover. It can be positive or negative depending on the pair, the direction and the broker, and it changes over time. The example uses a charge of $0.60 per standard lot per night, an invented figure, so check what your own broker charges.
• Margin. The deposit the broker holds to keep a position open. It is the position value divided by the leverage. With the price at 1.1000 and leverage of 1:30, one standard lot needs about $3,666.67. It scales directly with the lot, so doubling the lot doubles the margin. The rules vary by broker and by country. See margin call and stop out.

Worked example: a ladder on a currency pair

Everything here is made up. The ladder starts at 0.01 lot and doubles after each stopped trade. Each trade has a 20 pip stop and a 20 pip target, and one pip on one standard lot is worth $10. The spread is 1.2 pips on every step. Each step is held one night. The account is $2,000.00. A stopped step costs the stop distance plus the spread, plus swap. A step that hits its target earns the target less the spread, less swap. The column "Net if it wins" is that win minus everything lost on the steps before it.

Calculated by us on 29 Oct 2026, exact. Read "Net if it wins" first. With no spread or swap, a win at step 1 would net $2.00. With them it nets $1.87, because $0.13 of the target went to costs. A win at step 5 nets -$1.91 and at step 6 -$5.94. The net does not stay at the starting size. It shrinks as the ladder grows, because the extra costs on the larger lots are paid on top of the earlier losses. The goal of "one winner repays everything plus a unit" falls short once spread and swap are counted, and the shortfall widens at each step.

Then read the margin columns. The account covers the margin for steps 1 to 6. At step 7 the platform needs $2,346.67 and only $1,866.06 of equity is left, so the order cannot be opened, and the bot sits holding a loss of $133.94 with no way to continue the ladder. The total lost passes the whole account at step 10. How a broker handles an account in that state varies, and some close positions automatically.

How the odds look, and why they do not fix it

Suppose a step is stopped with a chance of 50 percent, and the ladder is capped at 5 steps. Then a whole cycle ends at the cap with a chance of 0.5 to the power 5, which is 3.125 percent, about 1 in 32. A cycle that ends on step 1 nets $1.87. One that ends on step 5 nets -$1.91: the last trade was a winner and the cycle still lost money, because costs outgrew the target. At the cap it loses $65.91. The cap is an assumption you choose, and the 50 percent is an assumption about the market. If stops are hit more often than that, which is what a trend does, the cap is reached more often. Because the spread and swap lower every win, the average over cycles falls below the break-even point that a cost-free ladder would have.

Put plainly: the ladder converts many small wins and costs into a rare large loss. The expected value does not turn positive because the wins are frequent. See expectancy.

What this means for a bot

• A backtest may not show it. A run with no long losing streak looks smooth. Check the longest run of stops in the test and the lowest margin level. See MT4 strategy tester and overfitting.
• A cap sets the size of the loss, not its existence. A cap at step 5 here means a loss of $65.91, which is 3.3 percent of the account.
• Tests with a flat low spread flatter the late steps. Spread widens around news and rollover, when stops are most likely to be hit.
• Weekend gaps jump over stops. A stop can fill at a worse price than set, which makes a step lose more than the table shows. See slippage.

What this page does not tell you

It does not say that a ladder never wins, or that it cannot win small amounts for a long time. It does not know your broker's spread, swap or margin rule. It cannot say how often a long losing run happens on a given pair. It shows how the pieces fit and where the limit is.