The short answer
Grid trading is a systematic strategy that lays a lattice of orders at fixed intervals around a price, and whether it pays is decided entirely by whether the market ranges or trends, because the structure that banks every oscillation in a range is the same one that blows up in a trend. The screenshots show the first half of that sentence and hide the second, which is why the equity curve looks magical right up to the margin call.
I want to lead with that asymmetry because it is the whole story of grid trading condensed into one sentence. The strategy is not wrong, it is a range tool used as an all-weather tool, and the gap between those two uses is where the money disappears.
The honest framing is that grid trading is a legitimate systematic method that most retail traders deploy in the wrong market, and the trend risk it hides in unrealised drawdown is what ends most grid accounts.
The wider strategy context is in the trading strategies guide, and this page covers the range-trapping style at the systematic end of the field.
What grid trading actually is
A grid places a series of buy and sell orders at fixed price intervals, called the grid spacing, above and below a reference price. The classic neutral grid puts buy limit orders below the current price and sell limit orders above it, so the strategy buys as price falls and sells as price rises.
Each time price crosses a level, an order fills, and each time price crosses back, that position closes at a profit roughly equal to the grid spacing. The grid does not care about direction, it cares about movement, which is why it thrives on the back-and-forth of a range.
The spacing is the key design choice, because it sets both the profit per cycle and the number of levels the grid opens in a move. A tight spacing banks small profits often but opens many positions in a trend, while a wide spacing banks larger profits but fills rarely in a quiet market.
I think of a grid as a machine that rents out the market's chop, collecting a toll each time price crosses a line, and the rent is only positive when the market actually chops.
Why grid trading looks magical in a range
In a genuine range, grid trading produces the equity curve that every strategy seller dreams of, and the reason is mechanical. Price oscillates between a floor and a ceiling, and every oscillation triggers a buy at the low end and a sell at the high end, each one a small profit.
The equity curve rises in a near-straight line because the range converts price movement into captured profit regardless of which way the market breaks first. The range is the grid's ideal habitat, and in that habitat the strategy genuinely works.
This is the equity curve that gets screenshot and shared, and it is real, for as long as the range holds. The danger is not that the curve is fake, it is that the curve is conditional on a market state that does not last.
I have run grids on ranging pairs and watched the line climb for days, and the climb is exactly what makes the strategy seductive, because the good period feels like the strategy's true nature rather than a market condition it happened to suit.
Why grid trading blows up in a trend
The same structure that captures every oscillation in a range accumulates disaster in a trend, and the mechanism is the grid's defining weakness. In a sustained downtrend, the buy limit orders below price keep filling as price falls through them, and each one is immediately underwater.
The grid does not close those losing positions, it holds them open, because the strategy assumes price will oscillate back and let each one close at a profit. In a trend that assumption fails, and the losing positions pile up while the grid keeps opening new ones at lower levels.
The unrealised drawdown grows with every level the trend triggers, and it keeps growing as long as the trend runs, with no natural stop. The grid is short volatility without knowing it, and a trend is exactly the volatility event that unravels a short-volatility position.
I treat a grid as a bet that the market will not trend, because that is what it is, and a bet that the market will not trend is a bet that loses eventually in every market.
The maths of the blow-up
The blow-up is not a fluke, it is arithmetic, and running the numbers once is enough to cure the seduction. Suppose a grid uses one mini lot per level with a 20-pip spacing, and price trends 200 pips against it, which is an ordinary intraday move on a volatile pair.
| Levels triggered | Open positions | Avg adverse move | Cumulative floating loss |
|---|---|---|---|
| 5 levels (100 pips) | 5 | ~50 pips | ~$250 |
| 10 levels (200 pips) | 10 | ~100 pips | ~$1,000 |
| 15 levels (300 pips) | 15 | ~150 pips | ~$2,250 |
Read the table downwards and the trap is visible. The loss does not grow linearly with the trend, it accelerates, because each new level adds a fresh loser on top of the deepening average adverse move on every older level.
The cumulative floating loss is the number that ends the account, and it reaches a margin call on a modestly sized account well before the trend is anything unusual. A 300-pip move is a normal day on some pairs, and the grid above is already thousands of dollars underwater on it.
I run exactly this table before I deploy any grid, sized to my actual lot and spacing, because the table tells me how far price can run before the grid breaks, and that distance is always shorter than new grid traders assume.
With-trend versus against-trend grids
The neutral grid above is one variant, and the directional choice changes the risk profile entirely. An against-trend grid buys as price falls and sells as price rises, which profits in ranges but accumulates losers in trends, and it is the most common and most dangerous form.
A with-trend grid does the opposite, placing buy stops above price and sell stops below, so it adds positions in the direction of the move and profits in trends. The trade-off is that a with-trend grid bleeds in ranges, where it gets chopped into small losses as price whipsaws through its levels.
Neither variant solves the core problem, which is that a grid is a directional bet disguised as a market-neutral tool. The against-trend grid bets on mean reversion, the with-trend grid bets on momentum, and both lose when the market does the other thing.
I do not believe a grid can be made trend-safe by choosing its direction, because the direction is itself a forecast, and a forecast that is wrong produces the same drawdown a neutral grid produces in a trend.
The martingale variant, and why it kills accounts
The most lethal form of grid trading is the martingale variant, which doubles the lot size at each level as price moves against the position. The appeal is that a single retrace to the average closes every loser at breakeven plus a small profit, which makes the equity curve look unbeatable in a range.
The maths that produces the unbeatable curve in a range produces a near-certain blow-up in a trend, because doubling the lot at each level turns the accelerating-loss table above into an exponential one. Five levels against a martingale grid is not five losers, it is one plus two plus four plus eight plus sixteen lots of exposure, and the floating loss explodes.
The martingale grid is the strategy most often sold to beginners, because its backtest equity curve is the smoothest and its short-term win rate is the highest, and both properties come directly from the property that kills it. A strategy that wins every cycle until the one cycle that ends the account is a strategy with a hidden bomb, and the fuse is the first sustained trend.
I would not run a martingale grid on any account I wanted to keep, because the win rate is a fee the strategy charges for the right to eventually take the whole balance, and the eventually is always sooner than the backtest implies.
The cost weight grid trading carries
Grid trading is cost-heavy in a way that surprises traders who only see the gross equity curve, because a dense grid generates many fills and every fill pays the spread and the commission. The cost is the reason the net profit per cycle is smaller than the grid spacing suggests.
On a 10-pip grid spacing with a 0.8-pip spread and a 0.7-pip commission equivalent, the cost is 1.5 pips, which is 15 percent of the spacing before the position even closes in profit. A grid that banks 10 pips gross per cycle banks roughly 8.5 net, and that gap widens on tighter spacings where the cost is a larger fraction of each cycle.
The cost also compounds against the strategy in a trend, because the losing positions pay swap every night they sit open, and a grid stuck in a trend holds its losers for days or weeks. The swap bleed turns a recoverable drawdown into a structural one, because the financing cost drains the account even if price eventually returns.
I factor the spread, commission and swap into any grid test before I trust the equity curve, because a grid that wins gross and loses net is the most common kind, and the cost is the silent partner that decides whether the spacing is actually profitable.
The market grid trading actually suits
Grid trading has a real market it suits, and naming it honestly is more useful than pretending the strategy is all-weather. The ideal market is a genuine range, with high volatility inside the band and clear, durable boundaries that hold for the duration of the grid.
Ranging pairs during quiet sessions, certain crosses that mean-revert by nature, and markets trapped between well-tested support and resistance are the conditions where a grid earns its keep. The strategy is a tool for a specific job, and the job is harvesting chop.
The market grid trading does not suit is any market that can trend, which unfortunately includes most pairs most of the time. A grid deployed without confirming the range is a grid deployed in the wrong market, and the wrong market is where the blow-up lives.
I only consider a grid after I have confirmed a durable range on the higher timeframes, and the reading of support and resistance is what tells me whether the boundaries the grid relies on are actually there.
How to run a grid honestly
If you choose to run a grid, the honest version treats the trend risk as real and capped rather than ignored, and four rules keep it survivable. The first is a hard maximum drawdown stop that closes the whole grid at a predetermined loss, accepting that the trend sometimes wins.
The second is a capped total position size, so the grid never exceeds a fixed fraction of the account no matter how many levels trigger. The third is range confirmation before deployment, so the grid only runs in the market that suits it, and it is pulled the moment the range breaks.
The fourth is the rejection of martingale sizing, because fixed lots keep the blow-up arithmetic linear instead of exponential, and a linear loss is survivable where an exponential one is not. The method for sizing the capped risk is in the guide to volatility-based position sizing.
I run a grid only with all four rules, because a grid without them is not a strategy, it is an unbounded bet on the absence of a trend, and that bet is the one the market eventually calls.
Why most retail grid traders lose
The loss rate among retail grid traders is high, and the reasons are the same structural ones that drive the wider retail-loss data. ESMA reports 74% to 89% of retail accounts lose money, and grid trading concentrates several of the behaviours that figure reflects (ESMA).
The first reason is the trend blow-up above, which converts a long run of small wins into a single account-ending loss. The second is martingale sizing, which makes the blow-up exponential and near-certain on the first sustained move.
The third is over-leverage, because the grids that fit small accounts depend on the high-leverage, many-position combination that the loss data flags. The fourth is the cost drag, which turns a gross-winning grid into a net-losing one and erodes the account even between blow-ups.
I do not say grids cannot pay, because they can in the right range with the right rules, but the winners share a pattern of capped size, fixed lots, confirmed ranges, and a hard stop, and the losers share the absence of all four.
When to pick a different strategy
Grid trading is one tool, and most of the time a different tool fits the market better, which is the honest thing to say before anyone deploys a grid. In a trending market, a trend-following approach captures the move a grid would bleed to, and the framework is in the strategies guide.
In a market that ranges but where you cannot watch a grid, a simpler mean-reversion trade with a defined stop carries the range exposure without the unbounded trend risk. The point is to match the strategy to the market rather than to force one tool onto every condition.
The fastest styles have their own cost maths, and the comparison is in the guides to tick scalping and the 20 pips a day challenge. A grid is the right answer some of the time, and the rest of the time it is the wrong answer dressed up as a system.
I pick the strategy from the market I see, not the other way round, because forcing a grid onto a trend is the error that turns a legitimate range tool into the loss statistics above.
Common mistakes with grid trading
The mistakes that end grid accounts are predictable, and naming them is most of the defence. The first is running a grid without a hard stop, letting the trend blow-up run to margin call because the strategy assumes price will always return.
The second is martingale sizing, doubling lots per level, which turns a survivable linear loss into an exponential one. The third is deploying the grid without confirming the range, running a range tool in a market that is free to trend.
The fourth is ignoring costs, sizing the spacing on gross profit and discovering the net is negative only in live trading. The fifth is over-leveraging to fit the grid on a small account, which is the combination the loss data sits heaviest on.
I keep the defence to four rules, a hard stop, capped size, fixed lots, and a confirmed range, and most of the mistakes above fall away at one of those gates. A grid run inside those rules is a legitimate range strategy, and a grid run outside them is a blow-up waiting for its trend.