How Spot Grid works, and where it doesn't

Spot Grid is one of the easiest automations to explain and one of the easiest to misjudge. The mechanic takes a sentence: divide a price range into levels, buy at each level as price falls, sell at each level as price rises, repeat. The consequences take rather longer.

The mechanic

You define an upper bound, a lower bound and a number of grid levels. The bot divides the range and places limit orders at each level. When a buy fills, a corresponding sell is placed one level higher. When that sell fills, the pair is complete and the spread between the two levels — minus fees — is realised.

Nothing about this predicts direction. A grid is a mechanism for converting oscillation into realised trades. It performs best in a market that moves within a range and does so repeatedly.

The two parameters that matter most

• Range width — a wide range survives more movement but places levels far apart, so fills are infrequent. A narrow range fills often but is exited quickly when price trends.
• Level count — more levels mean smaller spreads per pair and more fee events. Fewer levels mean larger spreads and longer waits. The exchange fee schedule sets a hard floor on how tight a level spacing can be before the spread stops covering the cost of capturing it.

Where it doesn't work

Two situations degrade a grid, and both are ordinary market behaviour rather than edge cases.

When price falls below the lower bound, every buy level has filled and no sells remain to be triggered. The grid is fully allocated into an asset that is still declining, and it holds that inventory until price returns into range. The unrealised loss is a normal consequence of the design, not a malfunction — but it is a real loss, and it can be a large one.

When price rises above the upper bound, the opposite happens: every sell has filled, the position is fully in quote currency, and the bot sits idle while the asset continues appreciating without it. The grid has captured the range and forgone the trend.

Configuring with the exit in mind

1. 01Choose the range from observed behaviour, not hope — where has this pair actually traded over a period comparable to how long you intend to run it?
2. 02Decide in advance what you will do if the lower bound breaks. Holding inventory is a decision; make it deliberately rather than by default.
3. 03Check the level spread against the round-trip exchange fee. If the spread is not comfortably larger, the grid is working for the exchange.
4. 04Allocate on the assumption that the full lower half of the grid fills. That is the capital genuinely at risk.

Used with those constraints understood, a grid is a clean way to systematise repetitive spot activity. Used as a yield product, it eventually surprises the person running it.