A crypto insurance fund is a pooled reserve used under an exchange's rules to absorb eligible losses when liquidation closes leave a deficit beyond a trader's margin. Bankruptcy price identifies the zero-margin boundary or takeover accounting price; it is different from the liquidation trigger and the engine's actual close price. Compare those records, then check which reserve pool applies and what happens when its capacity or loss-management limits are reached.
The useful question is where the difference between takeover value and market execution goes. A favorable close can replenish a reserve; an unfavorable close can consume it. Neither outcome, by itself, tells you the trader's final wallet balance or guarantees that another position will remain open.
Liquidation price, bankruptcy price and close price
These three prices answer different questions:
| Price or reference | Its job | What it cannot establish alone |
|---|---|---|
| Liquidation trigger or account risk threshold | Determines when the venue intervenes | The price at which exposure is ultimately closed |
| Bankruptcy price | Identifies the applicable zero-margin or takeover boundary | A price available in the public order book |
| Engine's actual close price | Records the result of disposing of exposure | How the customer and reserve are each settled |
For its perpetual and expiry contracts, Bybit defines bankruptcy price as the point at which margin reaches zero and describes the liquidation engine taking over at that price. Its calculation depends on margin mode; cross-margin bankruptcy handling uses the account's broader risk state. See Bybit's bankruptcy-price documentation.
The distinction explains why a takeover price need not match a candle print. A chart records a specified market-price series, while a takeover can be an accounting transfer. Preserve each field's meaning instead of replacing the event record with the nearest chart price. The mark-price liquidation guide covers the separate question of which risk reference initiated the intervention.
Do not turn a bankruptcy-price formula into a liquidation calculator. Maintenance requirements can cause intervention before all margin disappears. Account activity can also change the relevant boundary between an earlier screenshot and the event.
Who receives a liquidation surplus?
Under Bybit's documented insurance-fund mechanism, the liquidated position settles at bankruptcy price. If the engine closes more favorably, the residual goes into the insurance fund; if it closes less favorably, the fund covers the excess loss. That is an allocation rule, not a promise to refund the difference to the liquidated trader. Bybit's insurance-fund explanation distinguishes the customer settlement from the market close.
The direction matters. Selling a long at a higher price is favorable. Buying back a short at a lower price is favorable. Calling every positive close-minus-bankruptcy difference a surplus would reverse the result for shorts.
For a deliberately simplified linear contract, let q be signed base-asset quantity: positive for a long, negative for a short. Let C be the engine close price and B the bankruptcy price. Before other costs or adjustments, the reserve contribution from that price difference is:
reserve contribution = q × (C − B)
A positive result adds to the modeled reserve; a negative result consumes it. This is a price-gap calculation in quote currency, not a complete exchange ledger or a formula for an account's trigger.
Worked example: two closes after the same takeover
Consider a hypothetical isolated-margin linear long representing 0.40 BTC, entered at 50,000 USDT/BTC with 1,000 USDT of margin. Assume no fees, funding, transfers or other positions. For this illustration, margin plus price PnL reaches zero at 47,500:
1,000 + 0.40 × (47,500 − 50,000) = 0 USDT
Assume intervention begins at 48,100, a separately supplied hypothetical trigger. At that reference, price PnL is −760 USDT and remaining modeled equity is 240 USDT. That remaining amount is not proof that a voluntary exit can still execute or that the engine will realize the same price.
Now assume the whole position is taken over at 47,500 under the settlement convention described above. Compare two mutually exclusive engine-close outcomes:
| Measure | Outcome A: better close | Outcome B: worse close |
|---|---|---|
| Engine close price, USDT/BTC | 47,800 | 47,100 |
| Price PnL measured from original entry | −880 USDT | −1,160 USDT |
| Original margin plus that market PnL | +120 USDT | −160 USDT |
| Price-gap contribution to reserve | +120 USDT | −160 USDT |
| Trader's modeled margin after bankruptcy-price settlement | 0 USDT | 0 USDT |
In Outcome A, 0.40 × (47,800 − 47,500) = 120 USDT replenishes the reserve. The customer's assumed takeover settlement still consumes the 1,000 USDT margin. Crediting the customer another 120 would apply a different allocation rule.
In Outcome B, 0.40 × (47,100 − 47,500) = −160 USDT. The market loss exceeds original margin by 160; the reserve must absorb that eligible gap under the assumed rules. Subtracting it again from the already-zero customer settlement would assign the same shortfall twice.
For a short, use negative quantity. A hypothetical −0.40 BTC position with bankruptcy price 52,500 and buyback price 52,800 produces −0.40 × 300 = −120 USDT: another deficit. The signs agree with the economic fact that an expensive buyback is worse for a short.
The examples exclude costs to isolate allocation. Actual charges need their own records, as the liquidation-fee ledger explains. Inverse contracts also require different payoff units; the linear versus inverse contract guide covers that boundary.
A headline reserve total hides the pool boundary
A balance is useful only after identifying the pool available to the contract. Bybit's insurance API includes settlement coin, pool balance, USD value and a symbols field identifying shared membership or a single-contract pool. Its stated refresh cadence is one minute for isolated pools and 24 hours for shared pools, with possible delivery delays during extreme volatility. Preserve the source update time alongside your retrieval time. Bybit's insurance-pool API documentation defines those fields and limits.
Here, an isolated insurance pool means a pool allocated to a contract. Do not confuse that reserve classification with a customer's isolated-margin setting.
Consider a second hypothetical worksheet. An applicable pool contains 900 USDT. Three eligible close deficits arrive: 160, 310 and 540 USDT. They total 1,010, exceeding the assumed resources by 110. Sequentially, the balance falls to 740, then 430, and cannot fund the final 540 in full. No contributions, recoveries or other adjustments are assumed.
A separate 20,000-USDT pool cannot be added merely because it appears on the same dashboard. You need the actual allocation or transfer rules. Conversely, a larger applicable pool does not establish how much loss is arriving concurrently. This worksheet tests resource arithmetic, not reserve adequacy or the probability of an event.
Reserve depletion is not the only possible limit
Avoid an operating rule that reads “positive fund balance means no ADL.” Bybit's current ADL-alert documentation includes contract-PnL drawdown conditions as well as pool-equity conditions. Some grouped-pool triggers therefore do not require the displayed balance to reach zero. Check the applicable condition and parameters rather than infer protection from the balance alone. Bybit's ADL-alert specification provides the venue-specific rules.
Selection and post-event exposure belong to a separate analysis. The auto-deleveraging guide explains how an opposing position can shrink and why the remaining hedge needs review.
Other venues can specify different residual-loss handling. Deribit's insurance-fund page describes covering bankrupt accounts' negative equity and, if that fund is exhausted, socializing further bankruptcies among that day's winning traders. It also identifies additional liquidation-fee proceeds as fund income. Those are Deribit's stated mechanics, not an extension of Bybit's takeover model. See Deribit's insurance-fund documentation.
Validate one event from records to allocation
Start with the account, venue entity, contract, settlement asset, margin mode and event time. Then work through four checks:
- Identify the recorded takeover or bankruptcy price and the quantity actually transferred. Keep any remaining position separate.
- Obtain the engine-close information where the venue supplies it. If unavailable, leave the realized reserve contribution unverified; a candle is insufficient evidence.
- Reconcile the customer settlement, commissions, event charges and adjustments without treating every record as an additional cash movement.
- Match the contract to its reserve pool and retain the dated fallback rules, membership and balance observations.
If the engine closes in pieces, weight the prices by quantity. For a hypothetical fixed 47,500 bankruptcy reference, 0.10 BTC closed at 47,800 and 0.30 BTC at 47,600 contribute 30 + 30 = 60 USDT. The quantity-weighted price is 47,650. An unweighted 47,700 average would incorrectly imply an 80-USDT contribution. Use each tranche's actual takeover reference if it changes.
When reconciling ordinary executions, the mixed maker/taker fill guide helps distinguish per-fill fees from order-level assumptions. Those fees still do not replace the takeover-allocation evidence.
Keep the product boundary explicit too. A spot-margin short has borrowed inventory and a repayment liability; the borrow-availability guide explains that separate lifecycle. A futures reserve description cannot establish how a spot loan is funded or repaid.
A useful completed record states the takeover value, known execution result, customer settlement, applicable pool and unresolved data. If the execution result is missing, retain that limitation. The next action is to obtain the settlement evidence or venue clarification, rather than manufacture a surplus or deficit from the chart.