Skip to main content

The platform model in one view

DerivaDEX is a hybrid derivatives exchange. It keeps custody and later settlement anchoring on Ethereum while moving sequencing, matching, and most live trading-state transitions into a faster off-chain path. That split exists because the system is solving several different problems at once:
  • protecting trader custody
  • admitting and sequencing requests with low latency
  • applying deterministic matching and risk rules
  • publishing account and order outcomes fast enough for trading clients
  • anchoring later settlement-sensitive state outside the immediate operator runtime

The system layers and their jobs

LayerPrimary jobWhat public docs expose
Wallet and smart contractsHold user-controlled collateral and execute on-chain actions such as deposits, withdrawals, and governance interactionsSmart Contract Addresses and Core Contract Reference
Request admissionVerify signatures, replay-window fields, and request validity before sequencingRequest Signing Reference, REST Authentication
Confidential submissionKeep signed request bytes private until the operator can process themRequest Encryption Reference
Matching and risk engineApply deterministic price-time matching, solvency checks, liquidation rules, and ADL fallbackTrading Safeties and Guards, Margin Requirements
Public data surfacesExpose snapshots, market data, and ordered account updates over REST and realtimeREST API, Realtime API, Order and Account Event Reference
Checkpoint and settlement layerAnchor evolving state later through checkpointed commitments on EthereumConsensus, Checkpointing, and Finality

What happens to one request

Read the lifecycle as six separate steps:
  1. A trader prepares a request that identifies the trader, strategy, intent, nonce, and replay-window fields.
  2. The trader signs that request with EIP-712 semantics and, for private write paths, encrypts the signed payload for operator submission.
  3. The operator validates admission requirements such as signer identity, supported market, tick size, notional limits, collateral sufficiency, and replay-window freshness.
  4. If admitted, the request enters sequencing and then matching, where price-time priority, solvency constraints, self-trade prevention, liquidation handling, and ADL fallback determine the live outcome.
  5. The result becomes visible through REST snapshots and realtime event streams such as ORDER_UPDATE, STRATEGY_UPDATE, and TRADER_UPDATE.
  6. Later, the evolving exchange state is anchored through checkpointed settlement processes on Ethereum.
The important point is that these steps do not happen on one clock. A request can be accepted for sequencing before it is fully executed, and execution can become visible before the later checkpoint/finality path matters for settlement-sensitive actions.

Why the architecture is split this way

DerivaDEX does not keep everything on-chain because perpetual trading depends on fast sequencing, frequent state changes, and bounded market microstructure. Pushing every step through blockchain timing would make execution quality, gas exposure, and visible pre-trade intent dominate the trading experience. It also does not keep everything off-chain because custody, later settlement truth, and governance-sensitive accountability would then rest too heavily on operator promises. The on-chain checkpoint layer is what keeps the public assurance story tied to Ethereum rather than to a private runtime alone. So the system deliberately separates:
  • fast-path trading mechanics
  • public data visibility
  • later settlement anchoring
That separation is the main idea readers need before the deeper explanation and reference pages make sense.

What users get from this model

ReaderMain benefitMain discipline required
TradersLow-latency order handling without per-trade gas costsUnderstand that mark price, margin, liquidation, and settlement timing are not the same thing
BuildersDeterministic request and event semantics across REST and realtimeSeparate sequencing receipts, execution outcomes, and later finality in client logic
Governance and compliance readersPublic safety controls plus an explicit settlement anchor outside the immediate operator runtimeRead explanation and reference together instead of looking for one monolithic policy page

What this page does not claim

This page explains the operating model. It does not replace:
  • exact request schemas or transport fields
  • exact error and retry behavior
  • exact matching, margin, liquidation, or ADL constants
  • legal, licensing, or private compliance guidance
Use the adjacent routes below when you need those facts.

Sources

Move to adjacent routes

Last modified on April 13, 2026