Mean Reversion Strategy: The Logic Behind Grid Expert Advisors
A mean reversion strategy assumes that price which has moved a long way from its recent average tends to come back towards it. Every grid expert advisor is that assumption turned into code, which is why a grid pays steadily for years and then has one terrible month.
The assumption
Mean reversion is the idea that an unusually stretched price tends to return towards its recent average. It is not a law, and nothing enforces it.
What a grid expert advisor does is convert that idea into a position that gets larger the further price moves away. The system is not predicting the return; it is paid more for it the longer it waits, and charged more for it if it never comes.
Why currency pairs
Equity indices drift upward over decades. A currency pair is a ratio between two economies, and the pairs a grid is usually run on spend most of their time inside a range that is wide but finite.
AUDCAD sits between two commodity economies, which is the kind of relationship a range-based rule is built for. Whether it stays that way is not something a record of the past can settle.
What the record shows
In the EURUSD Patient simulation over 2010 to September 2026 on Dukascopy data, 187 of 201 months closed positive — 93.0% — and all seventeen calendar years closed positive. Most baskets were short-lived: price came back quickly and the grid never became a grid at all.
That is what mean reversion looks like when it works, and also what makes it easy to oversell. A run of small wins is not evidence that the assumption is safe; it is evidence that the market was inside its range.
The interesting months are the negative ones, and they come in pairs. On the €10,000 reference account of the same simulation: April 2015 closed −€1,301 and May 2015 +€1,629; September 2014 closed −€709 and October +€1,044; July 2013 −€690 and August +€882.
That shape is the strategy’s whole character. The loss month is the market moving away while the baskets stay open; the recovery month is the shared target finally firing.
Nothing was realised in between — the loss existed the entire time as a floating loss on open positions.
The deepest floating loss of the whole simulation came later, −€2,788 on 21 August 2024, and the deepest fall from an equity peak was €2,788 over the same period.
How Kestrel applies this
Kestrel EA is a grid expert advisor for MetaTrader 4: it opens from its own signal, adds to the basket in small, capped steps when price moves against it, and closes the whole basket together at its target. It never closes a single trade at a loss; until the basket closes, the loss is floating.
Like the kestrel that hangs still in the wind, it waits, holds and closes once. The assumption held often enough over sixteen simulated years to produce sixteen positive years — and failed hard enough, on dated occasions, that the capital table matters more than the return table. See Kestrel Grid EA for MT4: Expert Advisor With Controlled Lots, the mechanics in Grid Trading Strategy: Rules, Risks and Lot Sizing, and the licence levels on Kestrel Forex Robot Price and Licence Levels.
Frequently asked questions
Is mean reversion the opposite of trend following?
In effect, yes. A trend follower buys strength and cuts quickly when it fails; a mean reversion system sells strength and holds while it continues. The first loses often and small, the second wins often and loses rarely and large.
Which market conditions break it?
A sustained directional move with no meaningful pullback. The system keeps adding against the move and the floating loss grows until price returns far enough for the shared exit. If it never returns, the position never closes on its own terms.
Does it work on every currency pair?
No. Pair behaviour, data quality and account sizing all change the result, and no historical record settles whether the assumption will hold in future.
Last updated: