Martingale is the idea that if you double the position after each loss, the next win recovers everything. It's mathematically true and practically ruinous, because the sequence that beats it — enough consecutive losses to exceed the account — is not rare, it is guaranteed given enough time. Yet most grid EAs use some form of lot progression, and many are perfectly usable. The difference is in the numbers, so here they are.
Exposure by level
Take a grid that adds a position every $10 Gold moves against it, starting at 0.01 lot.
Fixed lot (0.01 at every level): after 5 levels you hold 0.05 lot, average entry $20 from the first level. A further $10 adverse move costs 0.05 × $10 × 100 = $50.
Linear progression (+0.01 each level: 0.01, 0.02, 0.03, 0.04, 0.05): after 5 levels you hold 0.15 lot — three times the fixed-lot exposure. The same $10 move now costs $150.
Exponential (×2: 0.01, 0.02, 0.04, 0.08, 0.16): after 5 levels you hold 0.31 lot — six times fixed. The $10 move costs $310, and level 8 alone would be 1.28 lots.
Why the win rate is misleading
A martingale grid closes most baskets in profit, because most pullbacks end. So the history reads 95% winners, each small. The 5% — the trend that didn't end — carries a loss the size of all the winners combined, or larger. Average win $12, average loss $600, 95% win rate: expectancy is 0.95 × 12 − 0.05 × 600 = −$18.60 per basket. The equity curve climbs for months and then removes itself in a week. On a Myfxbook page this shows up as a huge gap between average win and average loss.
What makes lot progression survivable
- A maximum number of levels — the grid stops adding at, say, level 8, so the worst case is known in advance.
- A basket stop-loss — the whole basket closes at a defined floating loss (for example 15% of balance), turning an open-ended risk into a fixed one.
- Adaptive spacing — levels widen when volatility rises, so a trending day opens fewer positions.
- Linear rather than exponential progression, or a capped multiplier (×1.3, not ×2).
- Account sizing to the published max drawdown, not to the gain.
Sizing an account to a grid
Take the EA's live max drawdown from Myfxbook — say 32%. Assume the next one will be worse (they usually are): plan for 50%. If you want the account to survive that with margin to spare, run the EA at a lot size where the historical worst-case basket is at most 25–30% of your balance. In practice that's the minimum deposit on the product page, and it's why "$5,000 for Gold" isn't a marketing number — it's the drawdown math.
Where the Quantora robots sit
Smart Grid, Quantora EA and BB Grid use lot progression (linear or exponential, selectable) with a maximum level count, adaptive spacing and an optional basket stop-loss — the survivable kind, with the live drawdown published on each product page. Quantora Breakout has no progression at all: one position, fixed or risk-percent lot, hard stop. Pick by how much floating drawdown you can hold, not by the gain figure.
