What Matters
A trading outage does not necessarily close the position you cannot access. Your hedge may remain economically intact while you lose the ability to manage it. Closing the other leg or placing a replacement can then create directional exposure. DN’s Hedge Break Index models residual exposure across alternative position states, helping readers distinguish a confirmed broken hedge from an uncertain one.
DN Evidence Block
Last documentation review: 15 October 2026. This is a 2027 early edition, using information available in 2026.
Primary sources: Bybit’s reduce-only order guide; Hyperliquid’s exchange endpoint, order types and expired-action support documentation. These establish order-control mechanics, not comparative outage reliability.
Original DN contribution: a two-state residual-delta calculator, synthetic response scenarios and a state-reconciliation workflow.
Test period: public documentation reviewed on 15 October 2026. No funded trades, outage simulations or authenticated account tests. All prices, quantities, fee inputs and shocks are illustrative.
Research: Decentralised News. Independent reviewer: not assigned. No venues ranked or actively promoted.
DN Alpha Thesis: uncertainty can break the response before it breaks the hedge
Consider a matched book: long 2 BTC on one venue and short 2 BTC through a linear contract on another. Ignoring basis, funding and costs, the price sensitivities approximately offset. If the short venue’s interface becomes unavailable, the short does not disappear merely because the screen stops loading.
The trader now faces two questions: what exposure still exists, and what actions remain possible? Treating “cannot see it” as “no longer exists” can produce a duplicate hedge. Treating “last seen open” as “definitely still open” can leave a gap if the position was liquidated or closed meanwhile.
The response must therefore separate known positions, uncertain positions, pending orders and actual fills. This article measures that position-state uncertainty. It does not rank website uptime or estimate the probability of an outage.
Four events that need different responses
| Observed event | What it establishes | What it does not establish | Evidence needed |
|---|---|---|---|
| Interface or API unavailable | A control or visibility path is impaired | The position has closed | Official status, alternative authorised access and position records |
| Order request times out | The client lacks a definitive outcome | The order was rejected or never filled | Order ID, execution history and current position |
| Confirmed liquidation or closure | Some or all of the leg has changed | The other venue’s position changed too | Executed quantity, time, fees and remaining exposure |
| Withdrawal unavailable | Capital movement is constrained | Trading or hedging is also unavailable | Separate checks for order entry, risk controls and funding |
A platform can impair one path while others remain usable. Conversely, a working price feed does not establish that order entry or cancellation works. Use the component-level approach in DN’s existing resilience coverage, then reconcile the actual position state.
The response that looks protective can create exposure
Suppose the 2 BTC short remains open but you sell the accessible 2 BTC long. The aggregate position becomes short 2 BTC. If instead you keep the long and add another 2 BTC short elsewhere, the result is also a net short of 2 BTC while the original short survives.
If the original short has actually disappeared, either action can remove the original long exposure. The same action therefore has different consequences depending on the unknown position state. That is the central problem this tool exposes.
Unrealised gains on the inaccessible venue may also be unavailable to fund losses on the accessible venue. A book can retain near-zero aggregate delta and still suffer a local margin failure. Residual delta and funding availability must be assessed separately.
DN Hedge Break Index: measure the uncertain delta
This simplified model covers one underlying, a known long leg and an uncertain linear short leg. Enter the minimum and maximum percentages of the short that could still be open, then a signed change in exposure that you assume is fully executed elsewhere.
Residual quantity = known long quantity − original short quantity × retained fraction + signed adjustment.
Residual dollar delta = residual quantity × reference price.
DN Hedge Break Index = 100 × maximum absolute residual quantity across the two endpoint states / original short quantity.
Illustrative adverse-move loss plus action costs = maximum absolute residual quantity × reference price × assumed shock + action costs.
Action costs = absolute signed adjustment × reference price × (fee basis points + slippage basis points) / 10,000.
The index is a directional mismatch ratio, not a 0–100 safety rating. It can exceed 100. Its denominator is the original short leg, not account equity or the two legs’ combined gross notional. A zero value does not remove basis, custody, margin, execution or funding risk.
A negative adjustment reduces net delta: for example, selling spot or adding a short. A positive adjustment increases net delta. The tool captures the exposure change only; these routes have different collateral and execution requirements.
The $200,000 hedge: three illustrative responses
| Assumed executed adjustment | Residual if short is gone | Residual if short survives | Hedge Break Index | Adverse-move loss plus action costs |
|---|---|---|---|---|
| No change: 0 BTC | +2 BTC | 0 BTC | 100 | $10,000 |
| Reduce delta by 1 BTC | +1 BTC | −1 BTC | 50 | $5,300 |
| Reduce delta by 2 BTC | 0 BTC | −2 BTC | 100 | $10,600 |
The middle row has the smallest maximum absolute delta among these three examples. It is not a trade recommendation. It sacrifices neutrality in the surviving-short state, assumes the adjustment fills, and ignores liquidation, basis and account-specific constraints. Different known states or objectives can favour different actions.
If evidence instead confirms the short remains fully open, set both retention inputs to 100%. With no adjustment, residual delta is zero. If it confirms full closure, set both to 0%. The uncertainty interval should reflect evidence rather than a preferred answer.
DN Hedge-State Scenario Calculator
Use quantities in units of the same underlying. Enter the signed exposure adjustment as an executed outcome, not merely a submitted order. Defaults match the example above.
Order controls: useful, but local to the position they govern
Bybit’s reduce-only documentation describes instructions designed to reduce an existing position rather than open an unintended reverse position. Hyperliquid’s order-type documentation describes a similar purpose. These controls can prevent some local order mistakes; they do not make two venues execute atomically.
In particular, opening a replacement short on a third venue is a new position there. It cannot generally be expressed as a reduce-only instruction against a short held elsewhere. Permissions, collateral and supported instruments still have to be available.
Hyperliquid’s expired-action support page warns that repeated attempts during delays can result in multiple placements in the circumstances it describes. Its exchange API supports client order identifiers and order-status workflows. Identifiers help reconciliation; do not assume that any identifier automatically provides exactly-once execution or makes every retry safe.
Compare response paths before an incident
| Path | Potential use | Costs and access | Failure to consider |
|---|---|---|---|
| Wait while reconciling | Preserve the last known matched book | Funding, local margin and continued market movement | The uncertain leg may already be gone |
| Reduce the accessible long | Lower exposure if the short has closed | Execution fees, slippage and any loss of collateral recognition | A surviving short becomes less hedged |
| Add a short elsewhere | Replace confirmed missing short exposure | New margin, onboarding, permissions, fees and basis | A surviving original short creates excess short exposure |
| Partial adjustment | Explore intermediate exposure under uncertainty | Execution costs and residual risk in several states | Neither endpoint may remain neutral; this is not a universal solution |
These are scenario comparisons, not instructions to trade during an incident. The appropriate response depends on verified positions, mandate, liquidity, account health and the available controls.
The reconciliation record every multi-venue book needs
- Last confirmed state: quantities, contract type, margin mode, account and timestamp for each leg.
- Pending actions: order IDs, client IDs, acknowledgements, partial fills and cancellation requests.
- Independent checks: authorised position and execution-history queries, separate from a stale interface.
- Response assumptions: whether the inaccessible leg is confirmed open, confirmed closed or uncertain.
- Post-recovery cleanup: reconcile all fills and remaining orders before restoring normal automation.
A cancellation request is not a confirmed cancellation. An order acknowledgement is not necessarily a full fill. Automated retries should follow the venue’s documented semantics and reconcile the current state before adding exposure.
Connect the four DN risk questions
The Collateral Efficiency Index measures capital allocation. The Emergency Liquidity Index measures reserve arrival assumptions. The Venue Concentration Index measures shared dependency exposure. This tool adds a fourth question: what delta remains across plausible hedge states?
For strategy background, read Building a Multi-Exchange Hedged Position . For platform research, use DN Decision Briefs and Pathfinder. A second venue is useful only when its permissions, collateral and execution path are actually available.
Methodology, limitations and freshness
Version 0.1 uses first-order linear exposure for one underlying and two endpoint retention states. With fixed quantities, net delta is linear in retained short fraction, so the largest absolute residual occurs at an endpoint. The model assumes the known long and the executed adjustment remain unchanged during the shock.
It excludes inverse contracts, option gamma and vega, quanto exposure, cross-asset hedges, basis changes, funding, borrow costs, collateral repricing, partial execution of the adjustment, recovery values and liquidation mechanics. To examine an unfilled adjustment, rerun it at zero; to examine a partial fill, enter only that filled quantity. The model does not simulate the path while an adjustment is executing.
The shock is user-defined, not derived from volatility, outage duration or historical probabilities. The loss output is incremental to the reference state, includes one action’s modelled costs, and omits prior P&L and any later unwind. It is not an account solvency test.
Change log: v0.1, 15 October 2026: initial state-uncertainty framework, three synthetic response scenarios and calculator. Review when order semantics or API reconciliation procedures change. Submit corrections with primary evidence.
Primary sources
- Bybit: reduce-only order . Local position-reduction rules.
- Hyperliquid: order types . Reduce-only and execution-instruction definitions.
- Hyperliquid: exchange endpoint . Order request fields and client identifiers.
- Hyperliquid: expired actions . Specific delayed-action and repeated-order risks.
Frequently asked questions
Does a venue outage remove my hedge?
Not necessarily. The position may remain open while visibility or control is unavailable. Verify the actual position and execution history.
Can closing the accessible leg increase directional risk?
Yes. If the inaccessible opposing leg remains open, closing the accessible leg can leave a directional position.
Does reduce-only coordinate a hedge across exchanges?
No. It constrains the relevant local position under the venue’s rules. It does not guarantee simultaneous execution across venues.
Is the Hedge Break Index a failure probability?
No. It is a residual directional-exposure ratio under user-defined position states. It can exceed 100 and is not a safety rating.
Disclosure
Educational research, not personalised financial advice or an instruction to trade. Hedging can fail, and additional positions can increase losses and margin requirements. All examples are synthetic. Named platforms are documentation examples, not recommendations or verified LIVE endorsements. External source links are neutral and contain no referral codes. DN may earn referral compensation elsewhere on the site.
