When price leaves a fixed spot grid's range, new grid orders typically pause until it returns, while existing holdings remain in the account. Below the lower bound, the bot can hold accumulated base tokens; above the upper bound, it can hold mostly quote cash after sales. The range boundary is not an automatic stop loss. Check actual inventory, pending orders and exit settings before waiting, changing the range or restarting.
Bybit's Spot Grid FAQ documents this behavior for its fixed range: no new orders are placed outside the bounds until price returns. This guide considers an unleveraged spot grid without trailing or triggered exits. A particular bot's controls can change its behavior, so read those settings alongside the range.
The boundary limits order placement, not the asset price
The lower bound limits how far down the configured purchase ladder extends. Once eligible buys have filled on a decline, the bot can be left holding the resulting tokens while the market continues downward. Its software may be waiting, yet the economic exposure remains active.
At the upper bound, a rising market can exhaust the inventory assigned to sales. A further rally then produces no additional grid sales. Quote cash remains available for future purchases inside the old range, but it does not participate in the asset's additional rise.
Neither description proves that every order filled. Partial fills, rejected replacements, fee deductions and starting balances can leave a different mix. Inspect the holdings rather than concluding that a lower break means “100% base” or an upper break means “100% cash.”
Separate the range from any stop-loss or take-profit instruction. Bybit's setup guide describes optional price exits that terminate the bot and convert base tokens, with its stop-loss below the lower bound and take-profit above the upper bound. Those are additional actions. A boundary by itself does not instruct an exit.
Follow one grid in two opposite directions
This hypothetical teaching grid uses five price levels: $80, $90, $100, $110 and $120 per unit. At the starting $100 mark, it holds two units acquired at $100 each and $200 of quote cash. Starting equity is therefore $400.
Its initial orders buy one unit at $90 and one at $80, and sell one unit at $110 and one at $120. After a buy fills, it places a sale of that unit one level higher. After a sale fills, it places a purchase one level lower. Orders stay within the original range.
Every eligible order fills completely at its stated price before the next price event. There are no fees, spread, slippage, partial fills, borrowing, deposits or withdrawals. Assume one quote unit keeps a $1 value. Quote cash includes amounts reserved for accepted orders; an order reservation is not deducted from equity a second time. This is a small accounting model, rather than a reproduction of a venue's initialization algorithm.
Compare two separate paths from that same starting snapshot:
| Snapshot | Quote cash | Base units | Pending grid orders | Equity at the stated price |
|---|---|---|---|---|
| Start at $100 | $200 | 2 | Buy $80/$90; sell $110/$120 | $400 |
| Lower path reaches $80 after $90 | $30 | 4 | Sell $90/$100/$110/$120 | $350 |
| Lower path continues to $60 | $30 | 4 | Same four sales | $270 |
| Upper path reaches $120 after $110 | $430 | 0 | Buy $80/$90/$100/$110 | $430 |
| Upper path continues to $140 | $430 | 0 | Same four purchases | $430 |
On the lower path, cash becomes $200 − $90 − $80 = $30. Inventory becomes 2 + 1 + 1 = 4 units. At $60, equity is $30 + 4 × $60 = $270, which is $130 below the starting $400. There is no purchase below $80 to change that state.
The decline from $80 to $60 alone removes another 4 × $20 = $80 of marked value. The range has stopped expanding inventory in this model; it has not stopped the existing inventory losing value.
On the separate upper path, the two initial units sell for $110 and $120. Cash becomes $200 + $110 + $120 = $430, with zero units left. At $140, equity remains $430 under the unchanged quote-value assumption. The additional $20 rise after $120 creates no further account gain because the grid holds no base inventory.
The DCA versus grid inventory comparison explains how paired sales differ from a common position exit. Here, the operational question is what remains after the last eligible level has traded.
Returning inside the range does not undo the excursion
Suppose the lower path reverses from $60 and reaches $80 again. The model still holds four units and $30 cash, worth $350. Re-entering the range has not restored the original $400, and there is no immediate sale at $80. The next pending sale is at $90.
If that sale fills, cash becomes $120 and inventory falls to three units. Its paired $80 purchase is then reinstated. The next action follows the order ladder and actual fills, rather than the fact that an interface has switched from an out-of-range label to an active label.
A real bot may need to rebuild orders after a disconnect or a rejected replacement. Price returning inside the bounds is therefore a reason to verify resumed operation. Confirm which order should exist next and whether the exchange accepted it.
Also retain the original equity history. Binance's Spot Grid parameters distinguish matched grid profit from total profit, which also includes unrealized PnL. Resetting a dashboard or starting a new strategy record does not remove an earlier loss from the account.
Changing the range creates a new funding decision
Lowering the boundary can add purchases. In the $60 snapshot, the model has only $30 cash. Adding a one-unit buy at $60 needs $60, leaving a $30 funding gap if existing inventory is retained and no other funds are released.
If a hypothetical $30 deposit funds that purchase, the account then holds five units and no quote cash, worth $300 at $60. That is $30 more equity than before the deposit because $30 was contributed. Relative to the original $400 plus the new $30 contribution, the loss remains $130. The lower average purchase cost has not created value at the moment of the trade.
An upward restart can also change exposure. The upper path ends with $430 cash and no units at $140. A new grid may require an initial market purchase, creating fresh base exposure at the new price. Review the initialization preview before assuming a restart merely continues the old orders.
Provider details matter during edits. Bybit's setup guide says a range or grid-count modification can require additional investment or a changed base quantity. It also says modification cancels previous take-profit and stop-loss settings, which can be reconfigured during the change. Inspect the resulting protection as well as the new bounds. Bybit: Modify Spot Grid Bot Parameters.
Automatic trailing needs a separate check. Bybit's Trailing Up documentation describes shifting the grid when price reaches one interval above the upper limit, replacing the lowest buy with a buy at the previous upper limit. Its minimum-grid and balance conditions also apply. When that feature is enabled, inspect the current range and newly committed cash. The original upper bound alone no longer describes the ladder.
Use a decision worksheet before acting
Record the intended inventory after your action. “Restart the bot” is too vague to establish whether you intend to keep, sell or add tokens.
| Choice to evaluate | Record before the action | Verify afterward |
|---|---|---|
| Wait for a return into range | Current units, cash, exit settings and a review condition | Holdings remain within your approved exposure; exits are still configured |
| Stop automation and retain assets | Which pending orders should be canceled; which assets should remain | Terminal order states and the actual retained balances |
| Terminate and convert to quote | Expected sale quantity, settlement method and estimated costs | Actual fills, fees, residual base units and quote received |
| Edit or create a new range | Target units, extra funds, order sizes and new protection | New accepted orders, resulting inventory and active exit settings |
Stopping and settlement are separate choices. Bybit documents termination options that return the existing token mix or convert it into quote or base tokens. A conversion uses market prices and can incur trading fees. Read the selected settlement method before treating “terminated” as “holding cash.” Bybit: Grid Bot Settlement FAQ.
If inventory is leveraged or the bot uses futures, add position direction, collateral, funding and liquidation to this worksheet. The spot example does not model those obligations. The futures grid funding and leverage guide follows that separate ledger.
Reconcile the old orders before activating new ones
An edit or restart should begin from a known account state. Preserve the old bot identifier, exchange account, market, order IDs, latest fills and observation time. Check both pending orders and completed activity; an empty open-order list cannot explain how the inventory was acquired.
A cancellation can race with execution. Bybit's private order-stream documentation explicitly describes a concurrent cancel-and-fill case. Use cumulative executed quantities and final order states before sizing a new ladder. If the result remains unknown, apply the order timeout reconciliation process before another submission.
Then compare the new preview with the reconciled balances. Include base-token fee deductions, residual quantities and cash still reserved for other orders. A rounded order may fail a minimum or exceed the intended budget; the exchange order-minimum worksheet helps check whether the proposed ladder is feasible.
Finally, evaluate both a continued move away from the range and a return through it. Write the next expected order, maximum intended inventory, required cash and response to an unresolved cancellation for each path. The evidence guide for trading bots explains why reliable automation and evidence of a trading advantage require separate checks.
Before the next order, leave yourself a clear record: how many units remain, which orders can still fill, what a restart will buy or sell, and which condition will trigger the next review. That record makes an out-of-range decision inspectable even while the grid is waiting.