A grid EA places a ladder of buy and sell orders above and below price, then harvests the oscillation between them. It looks great in a ranging market and terrible in a trend. What most grid-EA buyers miss is that the broker, not the strategy, often decides which of those two outcomes they get.
The failure mode is specific. A grid holds many open positions at once, some of them deeply negative, sometimes for days. Three broker variables directly attack that structure: spread, position limits, and swap. Get any of them wrong and a grid that backtested beautifully collapses on a live account.
Why broker choice breaks grid EAs
- Spread is paid on every rung. A grid opens many positions. If each rung costs you an extra pip of spread, a grid that should break even at 15 pips of range now needs 20. Spread compounds with rung count, and this is invisible in most backtests.
- Position limits cap your ladder. Some brokers cap concurrent positions or total lots. A grid that needs 30 open orders hits a wall at 20 and silently stops adding rungs, which changes the whole risk profile.
- Swap bleeds you overnight. Grid positions often sit open for days. On a swap-heavy pair, the negative swap on the floating losers can exceed the grid’s harvested profit. Swap-free accounts fix this but usually hide the cost in a wider spread.
None of these show up on a broker’s homepage. They show up on your statement after two weeks.
The 4 requirements for a grid broker
- Raw spread with a commission, not a markup. For grids, raw + commission is almost always cheaper than a commission-free “standard” account, because you pay spread many times over. A 0.1 pip raw spread with a $7 round-turn commission beats a 0.9 pip no-commission spread once your grid has more than a handful of rungs.
- No low max-position cap. Check the maximum number of orders and the maximum open lots. You want headroom well above what your grid’s worst case requires.
- Execution reliability, not just speed. Grids care less about millisecond latency than scalpers and more about orders filling at the expected price during a fast move. Requotes or large slippage on a single rung misalign the whole ladder.
- Swap transparency. Know exactly what you pay to hold each lot overnight, on every pair your grid trades. If you need swap-free, verify how the broker prices it.
Top brokers compared
| Broker | Raw spread (EURUSD) | Commission | Grid notes |
|---|---|---|---|
| IC Markets | 0.0 pips | $7.00 round turn (MT) | No strategy restrictions, no min order distance, hedging allowed; the default for serious grids |
| Pepperstone | 0.0 pips | $7.00 round turn (MT) | Tier-1 regulation, reliable fills; Razor + cTrader if you want cBots |
| FP Markets | 0.0 pips | $6.00 round turn | Slightly cheaper commission, broad market access |
| Fusion Markets | 0.0 pips | $4.50 round turn | Cheapest commission here; offshore entity for most international clients |
The spread column is the floor, not the median. All four advertise “from 0.0”, but what you actually pay per rung is spread plus commission, and the spread widens at rollover and news. Backtest on the broker’s real historical feed before you commit.
Account type matters more than the broker
Every broker on the list has a “standard” account and a “raw/ECN” account. For a grid, always use raw. The standard account builds the spread into the price, which means you pay a wider spread on every single rung for the life of every position. A grid that opens 20 rungs and holds them overnight pays that markup 20 times plus swap. The commission on raw looks like a cost until you do the arithmetic.
The one exception is very small accounts. If your grid trades micro lots, a $7 round-turn commission per lot is proportionally expensive, and a tight standard account can be cheaper. Run the numbers for your actual lot size before deciding. I explain the raw-vs-standard tradeoff in detail in the IC Markets vs Pepperstone vs FP Markets comparison.
Minimum capital for a grid EA
There is no correct number, only a floor set by your grid’s worst case. Work it out backwards: take the maximum number of open rungs your grid can hold, the lot size of each, and the maximum adverse distance in pips, then compute the floating loss at the extreme. Add the margin required to hold all positions plus a buffer. If that number is bigger than your deposit, your grid is too big or your account is too small.
Most grid EAs that “blow accounts” were never given enough margin to survive their own worst case. The broker is rarely the cause; the sizing is.
How to test before going live
- Run the grid on a demo account at the exact broker and account type you plan to use.
- Test through a trend, not just a range. A grid that only ever saw ranges in backtest is untested.
- Log the spread on every rung for a week, including rollover and one news day.
- Compare swap charges against what you modeled. If they differ, find out why before funding.
I keep a running log of why I track every broker precisely because these numbers drift over time and between brokers.
Risk warning. Grid trading carries real account-wipe risk, especially when a strong trend runs against the grid and the floating drawdown exceeds your margin. A grid is a managed risk strategy, not a “set and forget” income machine. Size it so the worst trend you can imagine still leaves the account standing.