Grid vs martingale vs averaging: what is the difference?

In short: A grid says where orders go, averaging says why you add to a loser, martingale says how much bigger each new order is.

Grid, averaging and martingale are three different things. A grid is where orders sit: at fixed price steps. Averaging is adding to a losing position to move the average entry closer to the market. Martingale is how big each new order is: larger after a loss, classically double.

One basket can be all three at once. Below, one grid runs twice with the same step and the same number of orders; only the lot rule changes. The example: a cent account with $1,000, start lot 0.01, 11 orders, a $1.95 step, the first order at a gold price of $4,170, leverage 1:2000, stop-out at a 40% margin level. It is arithmetic of the settings, not a forecast or advice.

Are grid, averaging and martingale the same thing?

No. Grid, averaging and martingale are three different things: a grid is where orders sit, martingale is how big each order is, and averaging is what happens to the average entry price. The words travel together, but they answer three different questions:

• Grid: where? The price levels at which orders open.
• Martingale: how much? The rule that sets the lot of each order.
• Averaging: what happens to the entry price? Every order added below a losing buy pulls the average entry down, whatever its size.

A grid that keeps adding to one losing basket therefore averages by construction. Whether it is also a martingale depends only on the lot sequence. In betting, a martingale raises the stake after a finished loss; in a grid the earlier orders are still open and floating, so the basket gets heavier exactly while it is under water. Traders call that martingale too, loosely.

What do grid, averaging down and martingale mean?

In one sentence each: a grid is orders at planned price steps, averaging down is buying more after a fall to lower the average cost, and martingale is raising the size after each loss.

• Grid: orders placed at planned price distances, the step, so that more of them open as price moves. Here the basket closes together, so its break-even is the average entry price.
• Averaging (averaging down): buying more of an asset after its price has fallen, which lowers the average cost per unit: 100 shares at $50 plus 100 at $40 average $45. The same explainer warns that it raises risk, since more money rides on one falling position (The Motley Fool, 26 August 2018).
• Martingale: a betting strategy that raises the stake after each loss, so that one eventual win recovers all earlier losses and gains the original stake; classically it doubles. Wikipedia places it in 18th-century France. It says the strategy could be argued to win only with unbounded wealth, bets and time, because a limit on bet size or bankroll can end the sequence before the win (Wikipedia, "Martingale (betting system)", retrieved 8 October 2026).
• Anti-martingale (reverse martingale): the opposite lot rule: the size grows after wins and is cut after losses, not the other way round. It is a different lot rule, not a claim of lower risk.

The doubling arithmetic is short: eleven stakes of 1, 2, 4 … 1,024 units add up to 2,047 units. Make a unit 0.01 lot and that is 20.47 lots, the second column below.

Same step, two lot rules: what do the numbers say?

Depth is $19.50 in both columns and every order sits at the same level; only the lot of each order differs. Rebound is how far price must climb from the last order's level for the whole basket to sit at zero. Room is how much further price can fall below the last order before the broker's stop-out at a 40% margin level (see margin call vs stop-out). The $9,080 is more than twice the gold price itself, so on paper this buy basket runs out of price before it runs out of room.

The calculator keeps the multiplier at ×2 (the Midas default), so the equal-lot column uses the same formulas with the multiplier at 1.0.

Why does martingale bring break-even close, and why does that cost?

Martingale brings break-even close because doubling puts half of all the weight on the last order; it costs because each further dollar against the basket then costs $20.47 instead of $0.11, 186 times as much.

Think of the grid as a ladder: price levels are rungs, each order a weight hung on its rung. Break-even is the centre of gravity of the weights, the lot-weighted average entry (spread and costs ignored). Equal weights put it mid-ladder, $9.75 below the first order and half the depth. Doubling hangs half of all the weight on the last rung, so the centre slides to $17.56, only $1.94 above that rung: less than one $1.95 step.

That is the appeal. The cost sits in the other rows: the weight that pulls break-even toward the market also makes each further dollar cost $20.47 instead of $0.11, 186 times as much, and makes the loss when the 11th order opens 37 times bigger. Closer break-even, bigger exposure: the lot rule trades one for the other, and the number of orders is a second knob, and step, multiplier and order count covers both.

Exposure also runs into limits that ignore your balance: the 11th order is 10.24 lots, and a symbol has a maximal volume per deal and may have a cap on the combined volume of open positions and pending orders in one direction (MQL5 reference). Look up the values for your symbol and account.

Does a grid with constant lots avoid the martingale risk?

No. A grid with constant lots avoids the doubling, not the trend: the basket still averages down and nothing in the grid itself caps the loss. After the 11th order it is simply 0.11 lots long and each further dollar costs $0.11. With gold $500 below the first order (about 12%), the floating loss is $53.93 on $1,000: slow, but one-way. In this example neither grid has a stop: both rely on price coming back.

The $9,080 of room is large because 0.11 lots is a small position for $1,000, not because equal lots protect anything. Start the equal-lot grid at 1.86 lots (20.46 in total, about as heavy as the doubling basket) and the room after the last order is $38.3, with the $9.75 rebound still needed. What counts is the lots held against the balance, whichever rule piled them up; doubling just gets to a heavy basket faster, from a start that looks harmless.

What should you check in a grid EA to tell whether it is a martingale?

Read the lot sequence, not the label: divide the last lot by the first. 1.0 means constant lots, about 2.0 means doubling, and anything above 1.0 is a martingale-style size rule. Pages say "recovery", "progressive" or "non-martingale"; the lots of orders 1 to N say what the product does.

Five checks help:

• Divide the last lot by the first. 1.0 is constant, 2.0 is doubling, and anything above 1.0 is a martingale-style size rule, however mild.
• Ask what the lot depends on. A lot set as a share of the balance grows with the account, not with the losing orders, which alone is not a martingale. A lot that rises with the orders already open, or after a closed loss, is.
• Treat "non-martingale" as a claim to test. If the last order is larger than the first, the label means something narrower than it sounds (trading robot red flags).
• Look at the balance behind the first lot and the number of orders. The same lot sequence is a different risk on $1,000 than on $100,000.
• Ask whether there is any stop, and check the room to stop-out. Enter the lots, step and balance into the grid calculator; the room after the last order says more than the label.

Is Midas a grid EA, and is it a martingale?

Yes. Midas is a grid EA and it is martingale-style: every following order doubles from the start lot. Midas is an MT5 Expert Advisor for gold (XAUUSD) on hedging cent accounts. Its basket adds a larger order at each step: Risk, plainly calls it a doubling grid, and the panel settings say every following order doubles from the start lot. In this article's words, Midas is a grid, it averages, and it is martingale-style; the doubling column above uses its default settings. The optional lock is a different tool, an opposite order and not a size rule (locking a position). The 11 orders are walked through in martingale grid on gold; the terms are in the grid EA glossary.

What can go wrong

• A name does not change a loss. All three can end with the deposit gone: a doubling grid when price moves past its room, a constant-lot grid more slowly but with no cap of its own either.
• The calculator is kinder than the market. It assumes every order opens exactly at its level. Swap, commission, slippage and gaps are not in it, and spread is zero by default, so a real stop-out can come earlier.
• Your account is not the example. In MT5's forex formula, margin is lots × contract size ÷ leverage (MetaTrader 5 Help), and the stop-out level can be set in percent or in money (MQL5 reference), so read the one on your account. Rules differ by country: ESMA's 2018 package capped retail CFD leverage on gold at 20:1 and set close-out at 50% of the minimum required margin (ESMA, 1 June 2018); ESMA's own measures expired at the end of 31 July 2019 because most national authorities in the EU had adopted permanent rules of their own (ESMA). Check that your broker accepts clients from your country.
• No variant removes the risk. Wide room means a bigger move is needed, not that none is possible.

This is arithmetic of the settings, not a forecast. Trading with leverage carries risk: if the deposit is too small, a stop-out is possible. Size your deposit with a safety margin. Nothing here is individual investment advice.

Related: Martingale 2.0: why a doubling grid exits after a pullback of about $2 shows how the lot rule of a grid EA shortens the way back to break-even.

FAQ

What is the martingale strategy in trading?

In trading, martingale is a lot rule: the size grows after each losing step, doubling being the common case. It is distinct from a grid (where orders sit) and from averaging (adding to a loser). It brings break-even closer, but each further dollar against the basket then costs far more: 186 times as much at the 11th order in the example.

What is a grid EA, and how does grid trading work on gold?

A grid EA adds orders at set price steps as price moves against the first one and manages the whole basket, which closes together. On gold, each $1 of price changes the balance by lots times contract size; on a cent account with a 1-ounce contract, 1.00 lot moves $1. The lots held against the balance decide the risk: see the grid EA glossary and Risk, plainly.

What is a non-martingale grid EA, and does it avoid the risk of martingale?

One whose lot does not rise with each added order: constant lots, or a lot set from the balance alone. The label is not proof: read the lot sequence and divide the last lot by the first. It is still a grid that averages down, and a constant-lot grid can still lose the whole deposit if the lots are large for the balance.

Does martingale work in forex?

The lot rule alone creates no edge. Wikipedia says martingale could be argued to win only with unbounded wealth, bets and time, and that in a game with a negative expected value, bigger stakes only raise the average loss. A price chart is not a coin flip, so that does not settle trading either way. The table shows only a risk shape: a short way back to break-even, and each further dollar costing 186 times as much.

What is the best grid or martingale EA, and what should I check in one?

There is no ranking here, only criteria. Read the lot sequence and divide the last lot by the first, then check the balance behind the first lot, the number of orders, whether there is any stop, and the room to stop-out in the grid calculator. A label such as "constant lot" or "non-martingale" is not proof.

Is martingale or grid trading allowed by prop firms, and is it legal?

It depends on the current rules and terms of the prop firm or broker, which change, and on local law. Martingale is a lot-sizing rule (the lot grows after each losing step) and a grid is where orders sit; neither is a product. Read the current text or ask in writing. This page gives no legal advice.