You press Pause. The bot stops adding new activity, but the asset balance is still there. Has anything gone wrong?
Pausing a DCA bot, canceling its orders and closing its position are separate actions. A paused or stopped label does not prove that holdings were sold. Check the configured action, outstanding orders, actual fills and remaining exposure in the correct exchange account. A cancellation request can race with a fill, so its acknowledgment is not enough.
The purpose of a shutdown review is to establish a final account state. A button click starts that process; it does not necessarily finish it.
Write the state you want before choosing the command
Several different intentions can sound like “stop the bot.” You might want to prevent new deals while existing ones finish, stop further averaging while retaining take profit, cancel all outstanding orders, or exit the acquired asset immediately through the available market.
Those intentions imply different remaining exposures. Use explicit language:
| Intended state | Evidence needed |
|---|---|
| No new base deals | Entry rule disabled in the relevant scope |
| No more averaging purchases | Relevant buy orders final or canceled; conditional adds disabled |
| Existing position retained with exit management | Correct remaining quantity and intended exit orders |
| No strategy orders outstanding | Final status for every related order |
| No strategy exposure remaining | Reconciled inventory or position plus completed close fills |
A retained spot balance can be intentional. The problem is assuming it disappeared because software is no longer starting trades. Similarly, canceling take profit without selling leaves a holding with different exit management, not a closed position.
In KLINEO, consult the current DCA configuration guide. Its documented soft drawdown cooldown can cancel safety buys while keeping take profit active. Flatten-on-stop is a separate setting, and the interface may show Pause scheduled or Stop scheduled during reconciliation. Inspect the actual state rather than assigning one universal meaning to those labels.
Follow a cancellation race with actual quantities
Consider a fully hypothetical unleveraged spot position that owns 5 units bought for $500. A further purchase for 2 units at $80 is open. You decide to end the activity and submit a cancellation.
Before the cancellation becomes final, 0.4 units fill at $80. The other 1.6 units are then canceled. The account now owns 5.4 units with $532 of purchase cost before fees.
| Event | Filled inventory change | Remaining relevant quantity |
|---|---|---|
| Starting holding | 5 units already owned | 5 units |
| Cancellation requested for 2-unit buy | No final conclusion yet | Still uncertain |
| Partial buy executes | +0.4 units | 5.4 units owned |
| Remaining buy canceled | No reversal of the 0.4 fill | 5.4 units owned |
| First closing sale fills | −4 units at $79 | 1.4 units owned |
| Final closing sale fills | −1.4 units at $78 | 0 units in this construction |
A close sized from the old 5-unit snapshot could leave 0.4 units behind. A close request that itself only partly fills leaves another remainder. The relevant quantity comes from reconciled fills and inventory, not the originally intended ladder.
Bybit's cancel-order documentation explicitly describes cancellation acknowledgment as asynchronous and directs clients to confirm order status. That is a documented venue behavior; it does not establish how every exchange reports the same race.
Closing can realize a loss even when the command succeeds
Continue the invented example with the two closing fills in the table. Gross sale proceeds are:
4 × $79 + 1.4 × $78 = $425.20
Purchase cost was $500 + 0.4 × $80 = $532. The resulting loss is $106.80 before fees, spread, taxes or any omitted cost.
This does not estimate normal slippage or suggest that those prices are likely. It illustrates that successful operational completion and a profitable trade are different outcomes. A command can close the intended exposure and realize a substantial loss.
Conversely, a positive unrealized P&L at the moment you press Stop does not guarantee the proceeds of a later exit. An actual close needs executable liquidity and an accepted order. Read the stop-loss dollar-risk guide for the distinction between trigger price, quantity and realized loss.
No leverage or borrowing is included here. Futures add contract quantity, position mode, margin, funding and liquidation; do not infer a futures close from a spot asset-balance calculation.
Confirm orders before treating funds as free
Canceling a purchase can release quote funds that were locked for its unfilled remainder. It does not return quote spent on filled units. In the example, the 0.4-unit fill consumed $32 even though the rest of the order was canceled.
Likewise, canceling a sale can release asset units previously locked for it. That changes availability, not ownership. A free asset balance after cancellation can be evidence that the asset is still held.
Binance's user-data stream reference exposes order status and cumulative execution fields. Use the corresponding account and order evidence for the selected venue; an empty open-orders table alone misses already filled transactions and resulting inventory.
For a small residual quantity, minimum size and quantity-step constraints may prevent another ordinary sale order. The small-account minimums guide explains why such a residual should be recorded rather than rounded away in a claim that exposure is zero.
Disconnecting is not evidence of closure
Removing a connection or revoking credentials changes future access through that credential. It does not establish the disposition of earlier accepted orders, reverse executed trades or prove that holdings have been sold. Preserve a way to inspect the exchange account independently when ending a connection.
If a suspected credential incident requires immediate revocation, account verification still matters afterward. Treat connection security and position liquidation as separate outcomes to confirm. A disconnected dashboard can be missing the very observations needed to demonstrate that exposure is gone.
Unknown status is a state to resolve
A cancellation or closing request can time out. A missing response does not establish that nothing happened. Repeating the operation without checking can create duplicate actions, especially when the first request was accepted but its response was lost.
Identify the original order using its exchange or client identifier in the right account, market and environment. Compare status, cumulative executed units and any remaining order. Retain the identifiers and timestamps needed to explain the final state without exposing credentials.
The order-timeout reconciliation guide provides a fuller procedure. If a response is still ambiguous, report the uncertainty. Do not replace it with a confident “closed” label because the application stopped showing activity.
A connection error can also prevent the bot from seeing an order that remains active at the exchange. Use the venue's own account view as an additional observation when access is available. Logging out of the bot interface does not cancel accepted exchange instructions.
Keep take-profit protection in the review
If an exit order is already active, a close workflow must account for it. Otherwise one instruction may compete with another for the same asset quantity, or a later fill may arrive after your local inventory snapshot.
The examined KLINEO interface explains that flattening uses a live market sale after open sell orders are canceled. Pending protection reconciliation can delay a lifecycle change. This is a reason to inspect status and fills, not a promise of instant closure.
For a partial exit, check the remaining take-profit quantity and remaining holdings together. The take-profit accounting worksheet explains why a first profitable fill cannot be reported as though the whole position closed.
The same review applies to a grid with many paired orders. A canceled grid can leave asset inventory, while a completed pair may have recycled cash into another buy. The DCA-versus-grid ledger shows how those holdings can persist after a positive completed-trade result.
A useful shutdown record is small and specific
Record the bot and account scope, intended command, configuration version, outstanding order identifiers, final order statuses, filled quantities, remaining inventory and time of the last successful reconciliation. Distinguish “no strategy exposure observed” from “the entire account has no exposure,” especially if other bots or manual trades share it.
Do not change the pair, sizing or order logic while relying on a stale pre-change inventory. KLINEO's documented editing process requires pausing and clearing active exposure for material configuration changes; the current form and lifecycle messages should guide the operation.
A rehearsal should end with a quantity check, not merely a toast or a status color. If the task was only to suspend further buying, say that holdings remain. If it was to flatten, retain the evidence for every part of the close and any residual.
Stops, caps and shutdown software cannot guarantee a maximum loss through outages, gaps or rejected orders. The trading risk disclosure covers the broader limits of real execution.
Editorial method: This AI-assisted article combines current primary venue documentation, the examined KLINEO lifecycle messages and independently checked hypothetical fill accounting. It reports no live cancellation test, guaranteed shutdown timing or financial recommendation.