Grid trading EA: how a grid EA makes money, how it loses it all, and how to spot one

What a grid trading EA does

A grid trading EA places orders at fixed price intervals, say every 20 pips, above and below the current price or only in one direction. As price moves against the first position, the forex robot opens another, then another, each at a lower (or higher) level. When price comes back far enough that the whole basket is in profit by a set amount, it closes everything at once. Some grids take profit per level instead, but the defining feature is the same: multiple same-symbol positions stacked against an adverse move, closed as a group.

There is no prediction in a grid. It is a bet that price will come back before the account runs out of margin. On a ranging pair that bet wins for months, the trading bot books a tidy profit every few hours, and the MT4 or MT5 account looks like a cash machine.

The statistics you will see

A grid EA shows a win rate of 80–95% on closed trades, because positions are only closed when the basket is positive. The average win is small and remarkably uniform. Average loss, when it exists, is enormous, because a loss only happens when the basket is closed by a margin call or by the owner pulling the plug. Profit factor looks superb until that day, then falls below 1 in a single event.

The holding-time histogram has a signature shape: a tall cluster in the 15–60 minute and 1–4 hour buckets from quick basket closes, plus a long tail into 1–7 days and over 7 days from the positions that sat underwater waiting. Trades per week is high, often 50 or more, because every basket close produces several closed trades at the same second. The Z-score is strongly negative: wins arrive in clumps because a basket closes three, five or eight winners at once.

The two drawdown figures diverge. Max DD (balance) is tiny, maybe 3%. Max DD (equity) is where the truth lives: 30%, 50%, 70% floating loss while a basket waited. The equity curve on the growth chart shows deep sudden valleys that recover in a single step, a sawtooth hanging below a staircase. Open positions will regularly show three or more same-symbol trades in the same direction, with increasing floating loss the further they are from the current price.

A worked example of a grid blowing up

EURUSD, $5,000 account, a grid every 10 pips, 0.5 lots per level, basket take-profit of $50. Each level is $5 per pip. Price moves 80 pips against the first entry with no pullback, which EURUSD does most weeks. There are now eight positions open. The floating loss is 5 × (80 + 70 + 60 + 50 + 40 + 30 + 20 + 10) = $1,800, a 36% equity drawdown on an account whose balance curve still shows nothing but wins. Margin in use is 4 lots, around $4,300 at 1:100 leverage, so the broker is already close to refusing new orders.

Extend the move to 150 pips, which is a normal reaction to a central bank surprise. Fifteen levels, 7.5 lots, floating loss 5 × (10 + 20 + ... + 150) = $6,000. The account had $5,000. Every one of the $50 wins that came before, perhaps 120 of them for $6,000 total, has been given back in a single afternoon, and the account is gone. The grid did not fail unusually. It did what grids do.

Advantages, such as they are

• Works without any market forecast, which is why it is so easy to build and why free grid EA downloads are everywhere.
• On a genuinely range-bound cross with small lots and a huge account, a grid can grind out a few percent a year with modest equity dips.
• Produces a lot of closed trades quickly, which makes the statistics look mature before they are.

Disadvantages and failure modes

The failure mode is the one in the worked example and it is not a tail event; it is a scheduled event whose date is unknown. Every grid Expert Advisor without a hard basket stop will eventually meet a move larger than its margin can fund. The better grids cap the number of levels or close the basket at a defined equity loss. That makes the loss survivable but also makes it visible, which is why most vendors do not do it.

Secondary failures: swap on positions held for weeks, margin increases before weekends or news that force-close the basket at the worst moment, and hedged "recovery" baskets that double the exposure. Grids on gold are particularly lethal: a 150-pip move on EURUSD is 1.4%, and on gold 1.4% happens before lunch.

What the AI review and Score tend to flag

Expect the strategy type Grid, the grid watch flag, usually the no stop-loss high-severity flag since grid positions rarely carry individual stops, frequently the tail flag once any basket has been closed at a loss, and a deep drawdown watch flag if the equity drawdown has passed 40%. If the grid also increases lot size per level it will be classified as Martingale instead, because that check runs first.

The Score can be high for a while. Profitability is strong on profit factor, consistency is strong on green months. Risk control is where it falls: the drawdown input comes from the equity ledger, and chance of ruin from Risk Lab feeds in too. An account that has already taken its first big equity dip typically sits in the C or D range despite a 90% win rate.

Red flags when buying a grid EA

• Any vendor who says "no stop-loss needed, the grid recovers itself". That sentence is the failure mode.
• Balance drawdown under 5% with equity drawdown over 25% in Glance.
• No defined maximum number of levels or basket equity stop in the documentation.
• A record shorter than a year, which probably has not met its 150-pip move yet.
• Recommended minimum deposit under $1,000 with 0.1-lot levels. Do the margin arithmetic above.

How to check on an account page

Glance first: Max DD (equity) against Max DD (balance). Then the AI review for the grid score, the stop-loss percentage and the flags. Open positions is the live confession: count same-symbol, same-direction positions and read the floating loss. In Trade statistics, a win rate above 85% with an average win that barely varies confirms basket closes. In Trade records, sort by close time and look for clusters of trades closing at the same second. Risk Lab's chance of ruin ladder is less reliable for a grid because Monte Carlo shuffles closed trades, and the danger is in the open basket; treat the −50% and −80% rungs as lower bounds, not estimates. Performance depth's deepest drawdowns table will show depth and duration of the equity valleys if the account has been through one. If the account runs a grid alongside something else, By magic number isolates it. Read maximum drawdown explained, and compare the lot-progression variant in martingale EA.