Spot Grid Bots: What the Range and Grid Count Actually Do

A ladder, and the arithmetic behind it

A spot grid bot is a set of limit orders arranged between two prices: buys below the current price, sells above it, and after each fill a new order is placed one step away on the other side. The unit of profit is one matched pair — a buy filled at one grid line and a sell filled at the next one up. Binance’s guide calls that “grid profit” and defines it as “the realized profit of filled grid orders that are matched by one buy-order and one sell-order,” adding a sentence that matters for every calculation below: the trading fees incurred during the strategy are already deducted from it.

Two settings decide the whole strategy: where the range sits, and how many steps subdivide it. Everything else — the base asset, the amount per grid, the stop triggers — is downstream of those two.

Range: the boundary that suspends rather than closes

The guide is unusually clear about what the bounds do. The lower price is the bottom of the configured range, and “the system will no longer execute orders when the market price falls below the Lower Price.” The upper price works the same way in the other direction.

What happens next is the part worth reading twice. Using a BTC/USDC example with a 40,000 to 60,000 range, the guide says: “If the BTC price rises above 60,000 USDC or falls below 40,000 USDC, the strategy will be suspended.” And then: “When the price returns to the set price range, the strategy will resume.”

Suspended is not the same as closed. The bot stops trading and keeps its inventory exactly where it is. If price has left upward, that inventory is a position that is no longer being trimmed by sells. If price has left downward, the bot has been buying into a fall and stopped before the bottom. Either way, the range break converts a trading strategy into a holding, and the holding is marked to a price outside the range.

The optional stop loss and take profit are the only mechanisms that end a strategy for good. The guide constrains their placement — stop loss below the lower price, last price and trigger price; take profit above the upper price, last price and trigger price — which is a way of saying they must sit outside the range rather than inside it.

Grid count: spacing, and why the number is a fee decision

The grid count converts a wide interval into tradeable steps, and the two spacing modes do it differently.

Arithmetic mode: “Each grid has an equal price difference,” and the guide gives the formula directly.

Spread = (Upper Price - Lower Price) / Number of grids

Prices then step upward by that fixed amount: the first line is the lower price, the second is the lower price plus one spread, and so on.

Geometric mode: “Each grid has an equal price difference ratio,” with

Ratio = (Upper Price / Lower Price) ^ (1/n) × 100%

where n is the number of grids. The guide adds the consequence: in geometric mode, “the profits are fixed between each grid.”

That sentence is the argument for geometric mode in most crypto pairs. In arithmetic mode, a fixed dollar step is a large percentage move at the bottom of the range and a small one at the top, so the profit per matched pair varies with where the fill happened. In geometric mode every rung is the same percentage, which is what makes a per-grid profit comparable to a per-trade fee.

The fee test, worked

The guide never says “check that your per-grid profit exceeds your round-trip fee,” but the arithmetic does it for you in one line. A round trip in a grid is a buy and a sell, so it pays two fees. If the percentage step between rungs is smaller than that, the grid profit is negative before the position has moved at all — and because Binance reports grid profit net of fees, the interface will show a negative number rather than a small positive one.

Take a range of 40,000 to 60,000 — a ratio of 1.5 — and vary the grid count. Using geometric spacing:

The illustrative 0.10% round trip is an assumption, not a quote: it is 0.05% each way, and a reader should replace it with their own tier from the fee schedule, which is what Binance spot fees explains how to find. A maker order that rests on the book usually pays less than a taker order that crosses, and a grid bot’s orders are designed to rest — which is why the fee line is not as bad as the taker rate suggests, and why a badly configured grid can still cross the spread on every fill.

The general rule that falls out of the table: more grids is not more profit, it is more trades at a smaller margin. The total profit is the number of matched pairs multiplied by the profit per pair, and the number of matched pairs depends on how much the price oscillates, not on how many rungs you drew.

The capital constraint nobody puts in the range field

Spacing is not the only limit on grid count. Each rung needs an order, and each order must satisfy the exchange’s filters.

The public BTCUSDT symbol definition, read from Binance’s market data API on 2026-10-31, sets a minimum notional of 5 USDT, a quantity step of 0.00001 BTC and a tick size of 0.01 USDT. A grid with 500 rungs and 1,000 USDT of capital allocates 2 USDT per rung — below the notional minimum before any price movement. The bot will either refuse to start or place orders that cannot be filled.

Two consequences follow, and they pull in opposite directions:

• More grids need more capital to keep every rung above the minimum.
• More capital per rung does not increase the profit per matched pair — it increases the size of each pair, which multiplies both the profit and the loss.

The filter is the hard constraint and it is published. The softer constraint is that a rung whose order is worth exactly the minimum earns a profit so small that rounding in the fee calculation can erase it.

What the automatic parameters actually compute

Binance’s auto-parameter page is worth reading for what it does not contain. It gives formulas for the price bounds and no formula for the grid count.

The bounds come from a moving average and a standard deviation, with the multiplier fixed:

UB = Moving Average (MA) + Bollinger Bands (BBM) × Standard Deviation LB = MA − BBM × Standard Deviation

with BBM set to 3. The average price used is (Open + Close + High + Low) / 4 on one-hour candles, and the resulting band is then clamped to a range whose width depends on the lookback: minimum 1% × n^(1/3), maximum 15% × n^(1/3), default 5% × n^(1/3), where n is 7, 30 or 180 days.

Three standard deviations is a wide band by construction — it is designed to contain most of the recent price action, which is the same as saying it is designed to produce a range the price is unlikely to leave quickly. That is a reasonable default and it is not a forecast.

The grid count is left to the user, and the page’s other limits are the ones to remember: a minimum of 2 and a maximum of 500 grids for Spot Grid, 170 for the Trailing Up variant, and an automatic switch to a dynamic order mechanism above a default threshold of 170 orders. The auto-parameter page adds that you can create five grid strategies per trading pair.

What the strategy is actually exposed to

Three exposures sit outside the grid-profit number, and the interface reports them separately for a reason.

Unrealised P&L. The guide’s total-profit line is “Grid Profit + Unrealized P&L”. A bot that has earned 40 USDT of grid profit while holding a base asset that has fallen 200 USDT is a bot that is down. Grid profit measures the churn; unrealised P&L measures the inventory.

A directional move. Grids earn in oscillation and lose in trend. A sustained move in either direction converts the ladder into a one-sided position, and the range boundary is what stops it — by suspending the strategy with the position still on the books.

The assumption that the range holds. The bounds are set from history. Every parameter in the auto mode is computed from past candles, and none of them knows anything about the next week.

A checklist before switching one on

1. Write down the range and why. If the reason is “it looks like it has been ranging”, the range is a bet, not a measurement.
2. Compute the step: (upper / lower)^(1/n) − 1 for geometric spacing, or (upper − lower)/n for arithmetic.
3. Compare the step with your own round-trip fee, from your own fee tier, not from a blog post.
4. Check capital per rung against the minimum notional of 5 USDT on BTCUSDT and the quantity step.
5. Decide what happens at the boundary — nothing, a stop, or a manual review.
6. Read the total profit line, not the grid profit line. The difference is the position.

The order mechanics that decide whether a grid’s fills rest or cross are covered in maker versus taker orders, and the cost of the price you actually receive on each fill is in spread, depth and slippage.

Risk note. The formulas, limits, defaults and definitions in this article were read from Binance’s spot grid and auto-parameter guides, and the symbol filters from Binance’s public market data API, on 2026-10-31. The 0.10% round-trip figure is an illustration, not a quoted rate. A grid bot does not remove market risk; it redistributes it into inventory.