Grid and martingale scalping: how the maths ends

[Amanhasnoname]

Strategies that add to losing positions at set intervals can show beautiful equity curves. The mechanism explains both the appeal and the danger.

In a range, a grid collects small profits repeatedly. Price oscillates, and each swing closes some positions in profit. Martingale variants increase size after each loss, so one winning trade recovers everything plus a small gain.

The problem is the trend. When price moves far in one direction without a retracement, the open positions pile up, size grows exponentially, and margin runs out. The loss is typically larger than all the previous profits combined.

Backtests often miss it, because the sample period didn't contain a large enough move. Stress tests on the biggest historical trends, such as the franc shock or a prolonged dollar rally, usually expose it.

Survivors tend to run with strict caps on the number of levels, hard stops and small initial size.

The fundamental issue is a payoff structure with many small gains and a rare, huge loss.

A defensive variant some use is a hard equity stop for the whole grid, set at a level that would hurt but not destroy the account. It does not make the structure safe, but it turns a possible wipeout into a defined, survivable loss.

Has anyone tested a grid system against a long one-way trend?