Grid EA vs Martingale EA: What the Difference Actually Is

These two words get used as if they meant the same thing. They do not, and the difference is the whole risk profile. A grid decides where to add. A martingale decides how big. Most EAs on sale today do both at once, and it is the multiplier, not the grid, that decides what happens on the day the market does not come back.

Last updated 18 September 2026

Two separate ideas

A grid is a rule about where. The EA opens a position, and if the market moves against it by some distance, it opens another, and another, at fixed intervals. The idea is that price spends most of its time oscillating, so an average entry price gets dragged toward the market and the whole set closes together on a modest retrace.

A martingale is a rule about how much. Each new position is larger than the last, classically double. The idea is that one winner at the larger size recovers the losers behind it.

You can have either without the other. A flat grid adds positions at the same size. A martingale can operate on consecutive trades with no grid at all. In practice most EAs sold today combine them, because together they produce the smooth equity curve that sells.

Why the combination looks so good, for a while

A grid with a multiplier wins very often. Most oscillations do come back, the basket closes, and the account records another small profit. Months of this produce an equity curve that rises in a nearly straight line, with a win rate that can exceed 90 per cent.

What that curve hides is the shape of the losses. This class of strategy does not lose a little, often. It loses a lot, rarely. A year of clean months says almost nothing about the tail, because the tail has not happened yet.

This is why a six month track record on a grid martingale is not evidence in the way that six months of a trend following system is. Judge it by what the worst case costs, not by how often it wins.

The arithmetic, in full

Take a multiplier of 1.7 from a starting lot of 0.01, which is what our own EA uses by default, and follow all twelve levels:

LevelLot at this levelTotal open
10.010.01
30.030.06
60.140.34
90.701.69
112.024.88
123.438.31

Read the last column rather than the middle one. After six levels the basket holds about a third of a lot: nothing. The last two levels add more than six lots between them.

That is the whole character of the strategy in one table. Nothing much is at stake for most of the move, and then almost everything is at stake very quickly. An account that can comfortably fund level nine may be unable to fund level eleven, and an EA that cannot open level eleven is left holding ten losing positions with no mechanism to recover them.

What a per position stop loss does and does not do

Nearly every grid EA has a stop loss setting, and nearly every buyer reads it as the maximum they can lose. It is not.

The stop applies to each position. A basket of twelve positions at increasing size, all stopped out, loses many multiples of that figure. The stop caps a trade; it does not cap a basket.

What does cap a basket is an account level limit: a rule measured against equity or balance, that closes everything and stops trading when a threshold is crossed. Two are worth having:

If an EA has neither, the true maximum loss is the account.

What to ask before buying any grid EA

  1. What is the multiplier, and how many levels? Those two numbers give you the table above. Work it out for the lot size you intend to use.
  2. What is the largest basket the backtest actually opened, in lots? Could your account have funded it at your leverage?
  3. Are there account level limits, and are they on by default? Many ship switched off.
  4. Does the test include a violent period? March 2020, or a central bank surprise. A test that starts in 2021 is a test of a calm market.
  5. What happens when the market gaps past the last level? Some EAs close the basket immediately; others simply hold it.

Where we stand

Forex Success AI is a grid with a 1.7 multiplier, capped at twelve levels. The table above is our own default configuration, not a competitor's.

What we do about it: the per position stop loss exists but we do not present it as the limit of your risk; a daily loss limit and a floating drawdown limit are built in; a gap past the last entry closes the whole basket rather than holding it; and the defaults are published rather than hidden, which is what makes the arithmetic above possible to do at all.

What we will not tell you is that it cannot lose. Anyone selling a grid martingale who says that is either not being straight with you or has not held one through a bad week.

If you are deciding what to run it on, EAs on prop firm accounts covers why a firm's drawdown rule and this strategy need careful sizing, and the lot size calculator starts from what the account can fund.

Common questions

What is the difference between a grid EA and a martingale EA?

A grid adds positions at fixed price intervals as the market moves against it, and can do so with every position the same size. A martingale multiplies the size of each new position, classically doubling it. They are independent ideas: you can have a flat grid, a martingale without a grid, or both together, which is what most commercial EAs do.

Is a martingale EA guaranteed to blow the account eventually?

Not guaranteed, but the risk is structural rather than accidental. The strategy wins often and loses rarely and largely, so a long run of winning months proves less than it appears to. What decides survival is whether the account can fund the worst sequence the market produces, and whether the EA has a hard limit that closes everything before that point.

How large does a 1.7x multiplier get?

From 0.01 lots, the twelfth level alone is about 3.43 lots and the whole basket is about 8.31 lots. The first six levels total about 0.24 lots, so nearly all of the exposure arrives in the last few steps. That is why the equity curve looks calm for months and then moves violently.

Does a stop loss on each position make a grid safe?

No, and this is the most common misunderstanding. A per position stop caps one trade, but a basket holds many positions at increasing size, so the total loss if every one is stopped is far larger than the number in the settings suggests. Only an account level limit, measured against equity, caps what a bad day can cost.

How should I judge a seller's backtest of a grid EA?

Look at maximum drawdown before profit, check the test covers at least one violent period such as March 2020, and see whether the modelling quality and spread are realistic. Then ask what the largest basket in the test was, in lots, and whether your account could have funded it.